(2 min) Most approaches to emotion analysis of social media, literature, news, and other domains focus exclusively on basic emotion categories as defined by Ekman or Plutchik. However, art (such as literature) enables engagement in a broader range of more complex and subtle emotions. These have been shown to also include mixed emotional responses. We consider emotions in poetry as they are elicited in the reader, rather than what is expressed in the text or intended by the author. Thus, we conceptualize a set of aesthetic emotions that are predictive of aesthetic appreciation in the reader, and allow the annotation of multiple labels per line to capture mixed emotions within their context. We evaluate this novel setting in an annotation experiment both with carefully trained experts and via crowdsourcing. Our annotation with experts leads to an acceptable agreement of kappa = .70, resulting in a consistent dataset for future large scale analysis. Finally, we conduct first emotion classification experiments based on BERT, showing that identifying aesthetic emotions is challenging in our data, with up to .52 F1-micro on the German subset. Data and resources are available at https://github.com/tnhaider/poetry-emotion

(2 min) The paper copes with the task of automatic assessment of second language proficiency from the language learners' spoken responses to test prompts. The task has significant relevance to the field of computer assisted language learning. The approach presented in the paper relies on two separate modules: (1) an automatic speech recognition system that yields text transcripts of the spoken interactions involved, and (2) a multiple classifier system based on deep learners that ranks the transcripts into proficiency classes. Different deep neural network architectures (both feed-forward and recurrent) are specialized over diverse representations of the texts in terms of: a reference grammar, the outcome of probabilistic language models, several word embeddings, and two bag-of-word models. Combination of the individual classifiers is realized either via a probabilistic pseudo-joint model, or via a neural mixture of experts. Using the data of the third Spoken CALL Shared Task challenge, the highest values to date were obtained in terms of three popular evaluation metrics.

(2 min) The objective of this paper is to develop predictive models to classify Brazilian legal proceedings in three possible classes of status: (i) archived proceedings, (ii) active proceedings, and (iii) suspended proceedings. This problem's resolution is intended to assist public and private institutions in managing large portfolios of legal proceedings, providing gains in scale and efficiency. In this paper, legal proceedings are made up of sequences of short texts called "motions." We combined several natural language processing (NLP) and machine learning techniques to solve the problem. Although working with Portuguese NLP, which can be challenging due to lack of resources, our approaches performed remarkably well in the classification task, achieving maximum accuracy of .93 and top average F1 Scores of .89 (macro) and .93 (weighted). Furthermore, we could extract and interpret the patterns learned by one of our models besides quantifying how those patterns relate to the classification task. The interpretability step is important among machine learning legal applications and gives us an exciting insight into how black-box models make decisions.

(2 min) End-to-end models with auto-regressive decoders have shown impressive results for automatic speech recognition (ASR). These models formulate the sequence-level probability as a product of the conditional probabilities of all individual tokens given their histories. However, the performance of locally normalised models can be sub-optimal because of factors such as exposure bias. Consequently, the model distribution differs from the underlying data distribution. In this paper, the residual energy-based model (R-EBM) is proposed to complement the auto-regressive ASR model to close the gap between the two distributions. Meanwhile, R-EBMs can also be regarded as utterance-level confidence estimators, which may benefit many downstream tasks. Experiments on a 100hr LibriSpeech dataset show that R-EBMs can reduce the word error rates (WERs) by 8.2%/6.7% while improving areas under precision-recall curves of confidence scores by 12.6%/28.4% on test-clean/test-other sets. Furthermore, on a state-of-the-art model using self-supervised learning (wav2vec 2.0), R-EBMs still significantly improves both the WER and confidence estimation performance.

(2 min) Code-Switching (CS) remains a challenge for Automatic Speech Recognition (ASR), especially character-based models. With the combined choice of characters from multiple languages, the outcome from character-based models suffers from phoneme duplication, resulting in language-inconsistent spellings. We propose Contextualized Connectionist Temporal Classification (CCTC) loss to encourage spelling consistencies of a character-based non-autoregressive ASR which allows for faster inference. The CCTC loss conditions the main prediction on the predicted contexts to ensure language consistency in the spellings. In contrast to existing CTC-based approaches, CCTC loss does not require frame-level alignments, since the context ground truth is obtained from the model's estimated path. Compared to the same model trained with regular CTC loss, our method consistently improved the ASR performance on both CS and monolingual corpora.

(2 min) This paper addresses a series of complex and unresolved issues in the historical phonology of West Iranian languages. The West Iranian languages (Persian, Kurdish, Balochi, and other languages) display a high degree of non-Lautgesetzlich behavior. Most of this irregularity is undoubtedly due to language contact; we argue, however, that an oversimplified view of the processes at work has prevailed in the literature on West Iranian dialectology, with specialists assuming that deviations from an expected outcome in a given non-Persian language are due to lexical borrowing from some chronological stage of Persian. It is demonstrated that this qualitative approach yields at times problematic conclusions stemming from the lack of explicit probabilistic inferences regarding the distribution of the data: Persian may not be the sole donor language; additionally, borrowing at the lexical level is not always the mechanism that introduces irregularity. In many cases, the possibility that West Iranian languages show different reflexes in different conditioning environments remains under-explored. We employ a novel Bayesian approach designed to overcome these problems and tease apart the different determinants of irregularity in patterns of West Iranian sound change. Our methodology allows us to provisionally resolve a number of outstanding questions in the literature on West Iranian dialectology concerning the dialectal affiliation of certain sound changes. We outline future directions for work of this sort.

(2 min) The attention module, which is a crucial component in Transformer, cannot scale efficiently to long sequences due to its quadratic complexity. Many works focus on approximating the dot-then-exponentiate softmax function in the original attention, leading to sub-quadratic or even linear-complexity Transformer architectures. However, we show that these methods cannot be applied to more powerful attention modules that go beyond the dot-then-exponentiate style, e.g., Transformers with relative positional encoding (RPE). Since in many state-of-the-art models, relative positional encoding is used as default, designing efficient Transformers that can incorporate RPE is appealing. In this paper, we propose a novel way to accelerate attention calculation for Transformers with RPE on top of the kernelized attention. Based upon the observation that relative positional encoding forms a Toeplitz matrix, we mathematically show that kernelized attention with RPE can be calculated efficiently using Fast Fourier Transform (FFT). With FFT, our method achieves $\mathcal{O}(n\log n)$ time complexity. Interestingly, we further demonstrate that properly using relative positional encoding can mitigate the training instability problem of vanilla kernelized attention. On a wide range of tasks, we empirically show that our models can be trained from scratch without any optimization issues. The learned model performs better than many efficient Transformer variants and is faster than standard Transformer in the long-sequence regime.

(2 min) Knowledge is acquired by humans through experience, and no boundary is set between the kinds of knowledge or skill levels we can achieve on different tasks at the same time. When it comes to Neural Networks, that is not the case, the major breakthroughs in the field are extremely task and domain specific. Vision and language are dealt with in separate manners, using separate methods and different datasets. In this work, we propose to use knowledge acquired by benchmark Vision Models which are trained on ImageNet to help a much smaller architecture learn to classify text. After transforming the textual data contained in the IMDB dataset to gray scale images. An analysis of different domains and the Transfer Learning method is carried out. Despite the challenge posed by the very different datasets, promising results are achieved. The main contribution of this work is a novel approach which links large pretrained models on both language and vision to achieve state-of-the-art results in different sub-fields from the original task. Without needing high compute capacity resources. Specifically, Sentiment Analysis is achieved after transferring knowledge between vision and language models. BERT embeddings are transformed into grayscale images, these images are then used as training examples for pretrained vision models such as VGG16 and ResNet Index Terms: Natural language, Vision, BERT, Transfer Learning, CNN, Domain Adaptation.

(2 min) We present a novel approach to the problem of text style transfer. Unlike previous approaches requiring style-labeled training data, our method makes use of readily-available unlabeled text by relying on the implicit connection in style between adjacent sentences, and uses labeled data only at inference time. We adapt T5 (Raffel et al., 2020), a strong pretrained text-to-text model, to extract a style vector from text and use it to condition the decoder to perform style transfer. As our label-free training results in a style vector space encoding many facets of style, we recast transfers as "targeted restyling" vector operations that adjust specific attributes of the input while preserving others. We demonstrate that training on unlabeled Amazon reviews data results in a model that is competitive on sentiment transfer, even compared to models trained fully on labeled data. Furthermore, applying our novel method to a diverse corpus of unlabeled web text results in a single model capable of transferring along multiple dimensions of style (dialect, emotiveness, formality, politeness, sentiment) despite no additional training and using only a handful of exemplars at inference time.

(2 min) Recently, the emergence of pre-trained models (PTMs) has brought natural language processing (NLP) to a new era. In this survey, we provide a comprehensive review of PTMs for NLP. We first briefly introduce language representation learning and its research progress. Then we systematically categorize existing PTMs based on a taxonomy with four perspectives. Next, we describe how to adapt the knowledge of PTMs to the downstream tasks. Finally, we outline some potential directions of PTMs for future research. This survey is purposed to be a hands-on guide for understanding, using, and developing PTMs for various NLP tasks.

(2 min) Lexically constrained machine translation allows the user to manipulate the output sentence by enforcing the presence or absence of certain words and phrases. Although current approaches can enforce terms to appear in the translation, they often struggle to make the constraint word form agree with the rest of the generated output. Our manual analysis shows that 46% of the errors in the output of a baseline constrained model for English to Czech translation are related to agreement. We investigate mechanisms to allow neural machine translation to infer the correct word inflection given lemmatized constraints. In particular, we focus on methods based on training the model with constraints provided as part of the input sequence. Our experiments on the English-Czech language pair show that this approach improves the translation of constrained terms in both automatic and manual evaluation by reducing errors in agreement. Our approach thus eliminates inflection errors, without introducing new errors or decreasing the overall quality of the translation.

(2 min) In this study, a natural language processing-based (NLP-based) method is proposed for the sector-wise automatic classification of ad texts created on online advertising platforms. Our data set consists of approximately 21,000 labeled advertising texts from 12 different sectors. In the study, the Bidirectional Encoder Representations from Transformers (BERT) model, which is a transformer-based language model that is recently used in fields such as text classification in the natural language processing literature, was used. The classification efficiencies obtained using a pre-trained BERT model for the Turkish language are shown in detail.

(2 min) We focus on exploring various approaches of Zero-Shot Learning (ZSL) and their explainability for a challenging yet important supervised learning task notorious for training data scarcity, i.e. Depression Symptoms Detection (DSD) from text. We start with a comprehensive synthesis of different components of our ZSL modeling and analysis of our ground truth samples and Depression symptom clues curation process with the help of a practicing clinician. We next analyze the accuracy of various state-of-the-art ZSL models and their potential enhancements for our task. Further, we sketch a framework for the use of ZSL for hierarchical text-based explanation mechanism, which we call, Syntax Tree-Guided Semantic Explanation (STEP). Finally, we summarize experiments from which we conclude that we can use ZSL models and achieve reasonable accuracy and explainability, measured by a proposed Explainability Index (EI). This work is, to our knowledge, the first work to exhaustively explore the efficacy of ZSL models for DSD task, both in terms of accuracy and explainability.

(2 min) Word embeddings (e.g., word2vec) have been applied successfully to eCommerce products through~\textit{prod2vec}. Inspired by the recent performance improvements on several NLP tasks brought by contextualized embeddings, we propose to transfer BERT-like architectures to eCommerce: our model -- ~\textit{Prod2BERT} -- is trained to generate representations of products through masked session modeling. Through extensive experiments over multiple shops, different tasks, and a range of design choices, we systematically compare the accuracy of~\textit{Prod2BERT} and~\textit{prod2vec} embeddings: while~\textit{Prod2BERT} is found to be superior in several scenarios, we highlight the importance of resources and hyperparameters in the best performing models. Finally, we provide guidelines to practitioners for training embeddings under a variety of computational and data constraints.

(2 min) The prominence of figurative language devices, such as sarcasm and irony, poses serious challenges for Arabic Sentiment Analysis (SA). While previous research works tackle SA and sarcasm detection separately, this paper introduces an end-to-end deep Multi-Task Learning (MTL) model, allowing knowledge interaction between the two tasks. Our MTL model's architecture consists of a Bidirectional Encoder Representation from Transformers (BERT) model, a multi-task attention interaction module, and two task classifiers. The overall obtained results show that our proposed model outperforms its single-task counterparts on both SA and sarcasm detection sub-tasks.

(2 min) Event extraction is a fundamental task for natural language processing. Finding the roles of event arguments like event participants is essential for event extraction. However, doing so for real-life event descriptions is challenging because an argument's role often varies in different contexts. While the relationship and interactions between multiple arguments are useful for settling the argument roles, such information is largely ignored by existing approaches. This paper presents a better approach for event extraction by explicitly utilizing the relationships of event arguments. We achieve this through a carefully designed task-oriented dialogue system. To model the argument relation, we employ reinforcement learning and incremental learning to extract multiple arguments via a multi-turned, iterative process. Our approach leverages knowledge of the already extracted arguments of the same sentence to determine the role of arguments that would be difficult to decide individually. It then uses the newly obtained information to improve the decisions of previously extracted arguments. This two-way feedback process allows us to exploit the argument relations to effectively settle argument roles, leading to better sentence understanding and event extraction. Experimental results show that our approach consistently outperforms seven state-of-the-art event extraction methods for the classification of events and argument role and argument identification.

(2 min) Dialect and standard language identification are crucial tasks for many Arabic natural language processing applications. In this paper, we present our deep learning-based system, submitted to the second NADI shared task for country-level and province-level identification of Modern Standard Arabic (MSA) and Dialectal Arabic (DA). The system is based on an end-to-end deep Multi-Task Learning (MTL) model to tackle both country-level and province-level MSA/DA identification. The latter MTL model consists of a shared Bidirectional Encoder Representation Transformers (BERT) encoder, two task-specific attention layers, and two classifiers. Our key idea is to leverage both the task-discriminative and the inter-task shared features for country and province MSA/DA identification. The obtained results show that our MTL model outperforms single-task models on most subtasks.

(2 min) Pre-trained language models achieves high performance on machine reading comprehension (MRC) tasks but the results are hard to explain. An appealing approach to make models explainable is to provide rationales for its decision. To facilitate supervised learning of human rationales, here we present PALRACE (Pruned And Labeled RACE), a new MRC dataset with human labeled rationales for 800 passages selected from the RACE dataset. We further classified the question to each passage into 6 types. Each passage was read by at least 26 participants, who labeled their rationales to answer the question. Besides, we conducted a rationale evaluation session in which participants were asked to answering the question solely based on labeled rationales, confirming that the labeled rationales were of high quality and can sufficiently support question answering.

(2 min) We present CharacterChat, a concept and chatbot to support writers in creating fictional characters. Concretely, writers progressively turn the bot into their imagined character through conversation. We iteratively developed CharacterChat in a user-centred approach, starting with a survey on character creation with writers (N=30), followed by two qualitative user studies (N=7 and N=8). Our prototype combines two modes: (1) Guided prompts help writers define character attributes (e.g. User: "Your name is Jane."), including suggestions for attributes (e.g. Bot: "What is my main motivation?") and values, realised as a rule-based system with a concept network. (2) Open conversation with the chatbot helps writers explore their character and get inspiration, realised with a language model that takes into account the defined character attributes. Our user studies reveal benefits particularly for early stages of character creation, and challenges due to limited conversational capabilities. We conclude with lessons learned and ideas for future work.

(2 min) In this paper, we propose a novel spoken-text-style conversion method that can simultaneously execute multiple style conversion modules such as punctuation restoration and disfluency deletion without preparing matched datasets. In practice, transcriptions generated by automatic speech recognition systems are not highly readable because they often include many disfluencies and do not include punctuation marks. To improve their readability, multiple spoken-text-style conversion modules that individually model a single conversion task are cascaded because matched datasets that simultaneously handle multiple conversion tasks are often unavailable. However, the cascading is unstable against the order of tasks because of the chain of conversion errors. Besides, the computation cost of the cascading must be higher than the single conversion. To execute multiple conversion tasks simultaneously without preparing matched datasets, our key idea is to distinguish individual conversion tasks using the on-off switch. In our proposed zero-shot joint modeling, we switch the individual tasks using multiple switching tokens, enabling us to utilize a zero-shot learning approach to executing simultaneous conversions. Our experiments on joint modeling of disfluency deletion and punctuation restoration demonstrate the effectiveness of our method.

(2 min) Conventional representation learning algorithms for knowledge graphs (KG) map each entity to a unique embedding vector. Such a shallow lookup results in a linear growth of memory consumption for storing the embedding matrix and incurs high computational costs when working with real-world KGs. Drawing parallels with subword tokenization commonly used in NLP, we explore the landscape of more parameter-efficient node embedding strategies with possibly sublinear memory requirements. To this end, we propose NodePiece, an anchor-based approach to learn a fixed-size entity vocabulary. In NodePiece, a vocabulary of subword/sub-entity units is constructed from anchor nodes in a graph with known relation types. Given such a fixed-size vocabulary, it is possible to bootstrap an encoding and embedding for any entity, including those unseen during training. Experiments show that NodePiece performs competitively in node classification, link prediction, and relation prediction tasks while retaining less than 10% of explicit nodes in a graph as anchors and often having 10x fewer parameters.

(2 min) Prior work has revealed that positive words occur more frequently than negative words in human expressions, which is typically attributed to positivity bias, a tendency for people to report positive views of reality. But what about the language used in negative reviews? Consistent with prior work, we show that English negative reviews tend to contain more positive words than negative words, using a variety of datasets. We reconcile this observation with prior findings on the pragmatics of negation, and show that negations are commonly associated with positive words in negative reviews. Furthermore, in negative reviews, the majority of sentences with positive words express negative opinions based on sentiment classifiers, indicating some form of negation.

(2 min) The growth rate in the amount of biomedical documents is staggering. Unlocking information trapped in these documents can enable researchers and practitioners to operate confidently in the information world. Biomedical NER, the task of recognising biomedical names, is usually employed as the first step of the NLP pipeline. Standard NER models, based on sequence tagging technique, are good at recognising short entity mentions in the generic domain. However, there are several open challenges of applying these models to recognise biomedical names: 1) Biomedical names may contain complex inner structure (discontinuity and overlapping) which cannot be recognised using standard sequence tagging technique; 2) The training of NER models usually requires large amount of labelled data, which are difficult to obtain in the biomedical domain; and, 3) Commonly used language representation models are pre-trained on generic data; a domain shift therefore exists between these models and target biomedical data. To deal with these challenges, we explore several research directions and make the following contributions: 1) we propose a transition-based NER model which can recognise discontinuous mentions; 2) We develop a cost-effective approach that nominates the suitable pre-training data; and, 3) We design several data augmentation methods for NER. Our contributions have obvious practical implications, especially when new biomedical applications are needed. Our proposed data augmentation methods can help the NER model achieve decent performance, requiring only a small amount of labelled data. Our investigation regarding selecting pre-training data can improve the model by incorporating language representation models, which are pre-trained using in-domain data. Finally, our proposed transition-based NER model can further improve the performance by recognising discontinuous mentions.

(2 min) Atomic clauses are fundamental text units for understanding complex sentences. Identifying the atomic sentences within complex sentences is important for applications such as summarization, argument mining, discourse analysis, discourse parsing, and question answering. Previous work mainly relies on rule-based methods dependent on parsing. We propose a new task to decompose each complex sentence into simple sentences derived from the tensed clauses in the source, and a novel problem formulation as a graph edit task. Our neural model learns to Accept, Break, Copy or Drop elements of a graph that combines word adjacency and grammatical dependencies. The full processing pipeline includes modules for graph construction, graph editing, and sentence generation from the output graph. We introduce DeSSE, a new dataset designed to train and evaluate complex sentence decomposition, and MinWiki, a subset of MinWikiSplit. ABCD achieves comparable performance as two parsing baselines on MinWiki. On DeSSE, which has a more even balance of complex sentence types, our model achieves higher accuracy on the number of atomic sentences than an encoder-decoder baseline. Results include a detailed error analysis.

(2 min) The focus of our paper is the identification and correction of non-word errors in OCR text. Such errors may be the result of incorrect insertion, deletion, or substitution of a character, or the transposition of two adjacent characters within a single word. Or, it can be the result of word boundary problems that lead to run-on errors and incorrect-split errors. The traditional N-gram correction methods can handle single-word errors effectively. However, they show limitations when dealing with split and merge errors. In this paper, we develop an unsupervised method that can handle both errors. The method we develop leads to a sizable improvement in the correction rates. This tutorial paper addresses very difficult word correction problems - namely incorrect run-on and split errors - and illustrates what needs to be considered when addressing such problems. We outline a possible approach and assess its success on a limited study.

(2 min) Commonsense reasoning is one of the key problems in natural language processing, but the relative scarcity of labeled data holds back the progress for languages other than English. Pretrained cross-lingual models are a source of powerful language-agnostic representations, yet their inherent reasoning capabilities are still actively studied. In this work, we design a simple approach to commonsense reasoning which trains a linear classifier with weights of multi-head attention as features. To evaluate this approach, we create a multilingual Winograd Schema corpus by processing several datasets from prior work within a standardized pipeline and measure cross-lingual generalization ability in terms of out-of-sample performance. The method performs competitively with recent supervised and unsupervised approaches for commonsense reasoning, even when applied to other languages in a zero-shot manner. Also, we demonstrate that most of the performance is given by the same small subset of attention heads for all studied languages, which provides evidence of universal reasoning capabilities in multilingual encoders.

(2 min) Recurrent Neural Networks (RNNs), more specifically their Long Short-Term Memory (LSTM) variants, have been widely used as a deep learning tool for tackling sequence-based learning tasks in text and speech. Training of such LSTM applications is computationally intensive due to the recurrent nature of hidden state computation that repeats for each time step. While sparsity in Deep Neural Nets has been widely seen as an opportunity for reducing computation time in both training and inference phases, the usage of non-ReLU activation in LSTM RNNs renders the opportunities for such dynamic sparsity associated with neuron activation and gradient values to be limited or non-existent. In this work, we identify dropout induced sparsity for LSTMs as a suitable mode of computation reduction. Dropout is a widely used regularization mechanism, which randomly drops computed neuron values during each iteration of training. We propose to structure dropout patterns, by dropping out the same set of physical neurons within a batch, resulting in column (row) level hidden state sparsity, which are well amenable to computation reduction at run-time in general-purpose SIMD hardware as well as systolic arrays. We conduct our experiments for three representative NLP tasks: language modelling on the PTB dataset, OpenNMT based machine translation using the IWSLT De-En and En-Vi datasets, and named entity recognition sequence labelling using the CoNLL-2003 shared task. We demonstrate that our proposed approach can be used to translate dropout-based computation reduction into reduced training time, with improvement ranging from 1.23x to 1.64x, without sacrificing the target metric.

(2 min) A growing effort in NLP aims to build datasets of human explanations. However, the term explanation encompasses a broad range of notions, each with different properties and ramifications. Our goal is to provide an overview of diverse types of explanations and human limitations, and discuss implications for collecting and using explanations in NLP. Inspired by prior work in psychology and cognitive sciences, we group existing human explanations in NLP into three categories: proximal mechanism, evidence, and procedure. These three types differ in nature and have implications for the resultant explanations. For instance, procedure is not considered explanations in psychology and connects with a rich body of work on learning from instructions. The diversity of explanations is further evidenced by proxy questions that are needed for annotators to interpret and answer open-ended why questions. Finally, explanations may require different, often deeper, understandings than predictions, which casts doubt on whether humans can provide useful explanations in some tasks.

(2 min) Image classification is an important task in various machine learning applications. In recent years, a number of classification methods based on quantum machine learning and different quantum image encoding techniques have been proposed. In this paper, we study the effect of three different quantum image encoding approaches on the performance of a convolution-inspired hybrid quantum-classical image classification algorithm called quanvolutional neural network (QNN). We furthermore examine the effect of variational - i.e. trainable - quantum circuits on the classification results. Our experiments indicate that some image encodings are better suited for variational circuits. However, our experiments show as well that there is not one best image encoding, but that the choice of the encoding depends on the specific constraints of the application.

(2 min) Generative adversarial networks (GANs) synthesize realistic images from random latent vectors. Although manipulating the latent vectors controls the synthesized outputs, editing real images with GANs suffers from i) time-consuming optimization for projecting real images to the latent vectors, ii) or inaccurate embedding through an encoder. We propose StyleMapGAN: the intermediate latent space has spatial dimensions, and a spatially variant modulation replaces AdaIN. It makes the embedding through an encoder more accurate than existing optimization-based methods while maintaining the properties of GANs. Experimental results demonstrate that our method significantly outperforms state-of-the-art models in various image manipulation tasks such as local editing and image interpolation. Last but not least, conventional editing methods on GANs are still valid on our StyleMapGAN. Source code is available at https://github.com/naver-ai/StyleMapGAN.

(2 min) Deep generative models have recently demonstrated the ability to synthesize photorealistic images of human faces with novel identities. A key challenge to the wide applicability of such techniques is to provide independent control over semantically meaningful parameters: appearance, head pose, face shape, and facial expressions. In this paper, we propose VariTex - to the best of our knowledge the first method that learns a variational latent feature space of neural face textures, which allows sampling of novel identities. We combine this generative model with a parametric face model and gain explicit control over head pose and facial expressions. To generate images of complete human heads, we propose an additive decoder that generates plausible additional details such as hair. A novel training scheme enforces a pose independent latent space and in consequence, allows learning of a one-to-many mapping between latent codes and pose-conditioned exterior regions. The resulting method can generate geometrically consistent images of novel identities allowing fine-grained control over head pose, face shape, and facial expressions, facilitating a broad range of downstream tasks, like sampling novel identities, re-posing, expression transfer, and more.

(2 min) Recent work has shown that convolutional neural network classifiers overly rely on texture at the expense of shape cues. We make a similar but different distinction between shape and local image cues, on the one hand, and global image statistics, on the other. Our method, called Permuted Adaptive Instance Normalization (pAdaIN), reduces the representation of global statistics in the hidden layers of image classifiers. pAdaIN samples a random permutation $\pi$ that rearranges the samples in a given batch. Adaptive Instance Normalization (AdaIN) is then applied between the activations of each (non-permuted) sample $i$ and the corresponding activations of the sample $\pi(i)$, thus swapping statistics between the samples of the batch. Since the global image statistics are distorted, this swapping procedure causes the network to rely on cues, such as shape or texture. By choosing the random permutation with probability $p$ and the identity permutation otherwise, one can control the effect's strength. With the correct choice of $p$, fixed apriori for all experiments and selected without considering test data, our method consistently outperforms baselines in multiple settings. In image classification, our method improves on both CIFAR100 and ImageNet using multiple architectures. In the setting of robustness, our method improves on both ImageNet-C and Cifar-100-C for multiple architectures. In the setting of domain adaptation and domain generalization, our method achieves state of the art results on the transfer learning task from GTAV to Cityscapes and on the PACS benchmark.

(2 min) We present a method for estimating neural scenes representations of objects given only a single image. The core of our method is the estimation of a geometric scaffold for the object and its use as a guide for the reconstruction of the underlying radiance field. Our formulation is based on a generative process that first maps a latent code to a voxelized shape, and then renders it to an image, with the object appearance being controlled by a second latent code. During inference, we optimize both the latent codes and the networks to fit a test image of a new object. The explicit disentanglement of shape and appearance allows our model to be fine-tuned given a single image. We can then render new views in a geometrically consistent manner and they represent faithfully the input object. Additionally, our method is able to generalize to images outside of the training domain (more realistic renderings and even real photographs). Finally, the inferred geometric scaffold is itself an accurate estimate of the object's 3D shape. We demonstrate in several experiments the effectiveness of our approach in both synthetic and real images.

(2 min) Despite the empirical success of neural architecture search (NAS) in deep learning applications, the optimality, reproducibility and cost of NAS schemes remain hard to assess. In this paper, we propose Generative Adversarial NAS (GA-NAS) with theoretically provable convergence guarantees, promoting stability and reproducibility in neural architecture search. Inspired by importance sampling, GA-NAS iteratively fits a generator to previously discovered top architectures, thus increasingly focusing on important parts of a large search space. Furthermore, we propose an efficient adversarial learning approach, where the generator is trained by reinforcement learning based on rewards provided by a discriminator, thus being able to explore the search space without evaluating a large number of architectures. Extensive experiments show that GA-NAS beats the best published results under several cases on three public NAS benchmarks. In the meantime, GA-NAS can handle ad-hoc search constraints and search spaces. We show that GA-NAS can be used to improve already optimized baselines found by other NAS methods, including EfficientNet and ProxylessNAS, in terms of ImageNet accuracy or the number of parameters, in their original search space.

(2 min) Recently, visual Transformer (ViT) and its following works abandon the convolution and exploit the self-attention operation, attaining a comparable or even higher accuracy than CNNs. More recently, MLP-Mixer abandons both the convolution and the self-attention operation, proposing an architecture containing only MLP layers. To achieve cross-patch communications, it devises an additional token-mixing MLP besides the channel-mixing MLP. It achieves promising results when training on an extremely large-scale dataset. But it cannot achieve as outstanding performance as its CNN and ViT counterparts when training on medium-scale datasets such as ImageNet1K and ImageNet21K. The performance drop of MLP-Mixer motivates us to rethink the token-mixing MLP. We discover that the token-mixing MLP is a variant of the depthwise convolution with a global reception field and spatial-specific configuration. But the global reception field and the spatial-specific property make token-mixing MLP prone to over-fitting. In this paper, we propose a novel pure MLP architecture, spatial-shift MLP (S$^2$-MLP). Different from MLP-Mixer, our S$^2$-MLP only contains channel-mixing MLP. We utilize a spatial-shift operation for communications between patches. It has a local reception field and is spatial-agnostic. It is parameter-free and efficient for computation. The proposed S$^2$-MLP attains higher recognition accuracy than MLP-Mixer when training on ImageNet-1K dataset. Meanwhile, S$^2$-MLP accomplishes as excellent performance as ViT on ImageNet-1K dataset with considerably simpler architecture and fewer FLOPs and parameters.

(2 min) While common image object detection tasks focus on bounding boxes or segmentation masks as object representations, we consider the problem of finding objects based on four arbitrary vertices. We propose a novel model, named TetraPackNet, to tackle this problem. TetraPackNet is based on CornerNet and uses similar algorithms and ideas. It is designated for applications requiring high-accuracy detection of regularly shaped objects, which is the case in the logistics use-case of packaging structure recognition. We evaluate our model on our specific real-world dataset for this use-case. Baselined against a previous solution, consisting of a Mask R-CNN model and suitable post-processing steps, TetraPackNet achieves superior results (9% higher in accuracy) in the sub-task of four-corner based transport unit side detection.

(2 min) In this work, we propose a no-reference video quality assessment method, aiming to achieve high-generalization capability in cross-content, -resolution and -frame rate quality prediction. In particular, we evaluate the quality of a video by learning effective feature representations in spatial-temporal domain. In the spatial domain, to tackle the resolution and content variations, we impose the Gaussian distribution constraints on the quality features. The unified distribution can significantly reduce the domain gap between different video samples, resulting in a more generalized quality feature representation. Along the temporal dimension, inspired by the mechanism of visual perception, we propose a pyramid temporal aggregation module by involving the short-term and long-term memory to aggregate the frame-level quality. Experiments show that our method outperforms the state-of-the-art methods on cross-dataset settings, and achieves comparable performance on intra-dataset configurations, demonstrating the high-generalization capability of the proposed method.

(2 min) The renaissance of artificial neural networks was catalysed by the success of classification models, tagged by the community with the broader term supervised learning. The extraordinary results gave rise to a hype loaded with ambitious promises and overstatements. Soon the community realised that the success owed much to the availability of thousands of labelled examples and supervised learning went, for many, from glory to shame: Some criticised deep learning as a whole and others proclaimed that the way forward had to be alternatives to supervised learning: predictive, unsupervised, semi-supervised and, more recently, self-supervised learning. However, all these seem brand names, rather than actual categories of a theoretically grounded taxonomy. Moreover, the call to banish supervised learning was motivated by the questionable claim that humans learn with little or no supervision and are capable of robust out-of-distribution generalisation. Here, we review insights about learning and supervision in nature, revisit the notion that learning and generalisation are not possible without supervision or inductive biases and argue that we will make better progress if we just call it by its name.

(2 min) Generative adversarial networks (GANs) have gained considerable attention owing to their ability to reproduce images. However, they can recreate training images faithfully despite image degradation in the form of blur, noise, and compression, generating similarly degraded images. To solve this problem, the recently proposed noise robust GAN (NR-GAN) provides a partial solution by demonstrating the ability to learn a clean image generator directly from noisy images using a two-generator model comprising image and noise generators. However, its application is limited to noise, which is relatively easy to decompose owing to its additive and reversible characteristics, and its application to irreversible image degradation, in the form of blur, compression, and combination of all, remains a challenge. To address these problems, we propose blur, noise, and compression robust GAN (BNCR-GAN) that can learn a clean image generator directly from degraded images without knowledge of degradation parameters (e.g., blur kernel types, noise amounts, or quality factor values). Inspired by NR-GAN, BNCR-GAN uses a multiple-generator model composed of image, blur-kernel, noise, and quality-factor generators. However, in contrast to NR-GAN, to address irreversible characteristics, we introduce masking architectures adjusting degradation strength values in a data-driven manner using bypasses before and after degradation. Furthermore, to suppress uncertainty caused by the combination of blur, noise, and compression, we introduce adaptive consistency losses imposing consistency between irreversible degradation processes according to the degradation strengths. We demonstrate the effectiveness of BNCR-GAN through large-scale comparative studies on CIFAR-10 and a generality analysis on FFHQ. In addition, we demonstrate the applicability of BNCR-GAN in image restoration.

(2 min) Current low-precision quantization algorithms often have the hidden cost of conversion back and forth from floating point to quantized integer values. This hidden cost limits the latency improvement realized by quantizing Neural Networks. To address this, we present HAWQV3, a novel mixed-precision integer-only quantization framework. The contributions of HAWQV3 are the following: (i) An integer-only inference where the entire computational graph is performed only with integer multiplication, addition, and bit shifting, without any floating point operations or even integer division; (ii) A novel hardware-aware mixed-precision quantization method where the bit-precision is calculated by solving an integer linear programming problem that balances the trade-off between model perturbation and other constraints, e.g., memory footprint and latency; (iii) Direct hardware deployment and open source contribution for 4-bit uniform/mixed-precision quantization in TVM, achieving an average speed up of $1.45\times$ for uniform 4-bit, as compared to uniform 8-bit for ResNet50 on T4 GPUs; and (iv) extensive evaluation of the proposed methods on ResNet18/50 and InceptionV3, for various model compression levels with/without mixed precision. For ResNet50, our INT8 quantization achieves an accuracy of $77.58\%$, which is $2.68\%$ higher than prior integer-only work, and our mixed-precision INT4/8 quantization can reduce INT8 latency by $23\%$ and still achieve $76.73\%$ accuracy. Our framework and the TVM implementation have been open sourced.

(2 min) A popular way to create detailed yet easily controllable 3D shapes is via procedural modeling, i.e. generating geometry using programs. Such programs consist of a series of instructions along with their associated parameter values. To fully realize the benefits of this representation, a shape program should be compact and only expose degrees of freedom that allow for meaningful manipulation of output geometry. One way to achieve this goal is to design higher-level macro operators that, when executed, expand into a series of commands from the base shape modeling language. However, manually authoring such macros, much like shape programs themselves, is difficult and largely restricted to domain experts. In this paper, we present ShapeMOD, an algorithm for automatically discovering macros that are useful across large datasets of 3D shape programs. ShapeMOD operates on shape programs expressed in an imperative, statement-based language. It is designed to discover macros that make programs more compact by minimizing the number of function calls and free parameters required to represent an input shape collection. We run ShapeMOD on multiple collections of programs expressed in a domain-specific language for 3D shape structures. We show that it automatically discovers a concise set of macros that abstract out common structural and parametric patterns that generalize over large shape collections. We also demonstrate that the macros found by ShapeMOD improve performance on downstream tasks including shape generative modeling and inferring programs from point clouds. Finally, we conduct a user study that indicates that ShapeMOD's discovered macros make interactive shape editing more efficient.

(2 min) Visual explanation methods have an important role in the prognosis of the patients where the annotated data is limited or unavailable. There have been several attempts to use gradient-based attribution methods to localize pathology from medical scans without using segmentation labels. This research direction has been impeded by the lack of robustness and reliability. These methods are highly sensitive to the network parameters. In this study, we introduce a robust visual explanation method to address this problem for medical applications. We provide an innovative visual explanation algorithm for general purpose and as an example application, we demonstrate its effectiveness for quantifying lesions in the lungs caused by the Covid-19 with high accuracy and robustness without using dense segmentation labels. This approach overcomes the drawbacks of commonly used Grad-CAM and its extended versions. The premise behind our proposed strategy is that the information flow is minimized while ensuring the classifier prediction stays similar. Our findings indicate that the bottleneck condition provides a more stable severity estimation than the similar attribution methods.

(2 min) Reflecting on the last few years, the biggest breakthroughs in deep reinforcement learning (RL) have been in the discrete action domain. Robotic manipulation, however, is inherently a continuous control environment, but these continuous control reinforcement learning algorithms often depend on actor-critic methods that are sample-inefficient and inherently difficult to train, due to the joint optimisation of the actor and critic. To that end, we explore how we can bring the stability of discrete action RL algorithms to the robot manipulation domain. We extend the recently released ARM algorithm, by replacing the continuous next-best pose agent with a discrete next-best pose agent. Discretisation of rotation is trivial given its bounded nature, while translation is inherently unbounded, making discretisation difficult. We formulate the translation prediction as the voxel prediction problem by discretising the 3D space; however, voxelisation of a large workspace is memory intensive and would not work with a high density of voxels, crucial to obtaining the resolution needed for robotic manipulation. We therefore propose to apply this voxel prediction in a coarse-to-fine manner by gradually increasing the resolution. In each step, we extract the highest valued voxel as the predicted location, which is then used as the centre of the higher-resolution voxelisation in the next step. This coarse-to-fine prediction is applied over several steps, giving a near-lossless prediction of the translation. We show that our new coarse-to-fine algorithm is able to accomplish RLBench tasks much more efficiently than the continuous control equivalent, and even train some real-world tasks, tabular rasa, in less than 7 minutes, with only 3 demonstrations. Moreover, we show that by moving to a voxel representation, we are able to easily incorporate observations from multiple cameras.

(2 min) Deep learning-based image inpainting algorithms have shown great performance via powerful learned prior from the numerous external natural images. However, they show unpleasant results on the test image whose distribution is far from the that of training images because their models are biased toward the training images. In this paper, we propose a simple image inpainting algorithm with test-time adaptation named AdaFill. Given a single out-of-distributed test image, our goal is to complete hole region more naturally than the pre-trained inpainting models. To achieve this goal, we treat remained valid regions of the test image as another training cues because natural images have strong internal similarities. From this test-time adaptation, our network can exploit externally learned image priors from the pre-trained features as well as the internal prior of the test image explicitly. Experimental results show that AdaFill outperforms other models on the various out-of-distribution test images. Furthermore, the model named ZeroFill, that are not pre-trained also sometimes outperforms the pre-trained models.

(2 min) Alongside the prevalence of mobile videos, the general public leans towards consuming vertical videos on hand-held devices. To revitalize the exposure of horizontal contents, we hereby set forth the exploration of automated horizontal-to-vertical (abbreviated as H2V) video conversion with our proposed H2V framework, accompanied by an accurately annotated H2V-142K dataset. Concretely, H2V framework integrates video shot boundary detection, subject selection and multi-object tracking to facilitate the subject-preserving conversion, wherein the key is subject selection. To achieve so, we propose a Rank-SS module that detects human objects, then selects the subject-to-preserve via exploiting location, appearance, and salient cues. Afterward, the framework automatically crops the video around the subject to produce vertical contents from horizontal sources. To build and evaluate our H2V framework, H2V-142K dataset is densely annotated with subject bounding boxes for 125 videos with 132K frames and 9,500 video covers, upon which we demonstrate superior subject selection performance comparing to traditional salient approaches, and exhibit promising horizontal-to-vertical conversion performance overall. By publicizing this dataset as well as our approach, we wish to pave the way for more valuable endeavors on the horizontal-to-vertical video conversion task.

(2 min) Neural Architecture Search (NAS) often trains and evaluates a large number of architectures. Recent predictor-based NAS approaches attempt to address such heavy computation costs with two key steps: sampling some architecture-performance pairs and fitting a proxy accuracy predictor. Given limited samples, these predictors, however, are far from accurate to locate top architectures due to the difficulty of fitting the huge search space. This paper reflects on a simple yet crucial question: if our final goal is to find the best architecture, do we really need to model the whole space well?. We propose a paradigm shift from fitting the whole architecture space using one strong predictor, to progressively fitting a search path towards the high-performance sub-space through a set of weaker predictors. As a key property of the proposed weak predictors, their probabilities of sampling better architectures keep increasing. Hence we only sample a few well-performed architectures guided by the previously learned predictor and estimate a new better weak predictor. This embarrassingly easy framework produces coarse-to-fine iteration to refine the ranking of sampling space gradually. Extensive experiments demonstrate that our method costs fewer samples to find top-performance architectures on NAS-Bench-101 and NAS-Bench-201, as well as achieves the state-of-the-art ImageNet performance on the NASNet search space. In particular, compared to state-of-the-art (SOTA) predictor-based NAS methods, WeakNAS outperforms all of them with notable margins, e.g., requiring at least 7.5x less samples to find global optimal on NAS-Bench-101; and WeakNAS can also absorb them for further performance boost. We further strike the new SOTA result of 81.3% in the ImageNet MobileNet Search Space. The code is available at https://github.com/VITA-Group/WeakNAS.

(2 min) Despite advances in deep learning, robustness under domain shift remains a major bottleneck in medical imaging settings. Findings on natural images suggest that deep neural models can show a strong textural bias when carrying out image classification tasks. In this thorough empirical study, we draw inspiration from findings on natural images and investigate ways in which addressing the textural bias phenomenon could bring up the robustness of deep segmentation models when applied to three-dimensional (3D) medical data. To achieve this, publicly available MRI scans from the Developing Human Connectome Project are used to study ways in which simulating textural noise can help train robust models in a complex semantic segmentation task. We contribute an extensive empirical investigation consisting of 176 experiments and illustrate how applying specific types of simulated textural noise prior to training can lead to texture invariant models, resulting in improved robustness when segmenting scans corrupted by previously unseen noise types and levels.

(2 min) Optical colonoscopy (OC), the most prevalent colon cancer screening tool, has a high miss rate due to a number of factors, including the geometry of the colon (haustral fold and sharp bends occlusions), endoscopist inexperience or fatigue, endoscope field of view, etc. We present a framework to visualize the missed regions per-frame during the colonoscopy, and provides a workable clinical solution. Specifically, we make use of 3D reconstructed virtual colonoscopy (VC) data and the insight that VC and OC share the same underlying geometry but differ in color, texture and specular reflections, embedded in the OC domain. A lossy unpaired image-to-image translation model is introduced with enforced shared latent space for OC and VC. This shared latent space captures the geometric information while deferring the color, texture, and specular information creation to additional Gaussian noise input. This additional noise input can be utilized to generate one-to-many mappings from VC to OC and OC to OC. The code, data and trained models will be released via our Computational Endoscopy Platform at https://github.com/nadeemlab/CEP.

(2 min) Biological systems perceive the world by simultaneously processing high-dimensional inputs from modalities as diverse as vision, audition, touch, proprioception, etc. The perception models used in deep learning on the other hand are designed for individual modalities, often relying on domain-specific assumptions such as the local grid structures exploited by virtually all existing vision models. These priors introduce helpful inductive biases, but also lock models to individual modalities. In this paper we introduce the Perceiver - a model that builds upon Transformers and hence makes few architectural assumptions about the relationship between its inputs, but that also scales to hundreds of thousands of inputs, like ConvNets. The model leverages an asymmetric attention mechanism to iteratively distill inputs into a tight latent bottleneck, allowing it to scale to handle very large inputs. We show that this architecture is competitive with or outperforms strong, specialized models on classification tasks across various modalities: images, point clouds, audio, video, and video+audio. The Perceiver obtains performance comparable to ResNet-50 and ViT on ImageNet without 2D convolutions by directly attending to 50,000 pixels. It is also competitive in all modalities in AudioSet.

(3 min) A LINE Bot System to diagnose rice diseases from actual paddy field images was developed and presented in this paper. It was easy-to-use and automatic system designed to help rice farmers improve the rice yield and quality. The targeted images were taken from the actual paddy environment without special sample preparation. We used a deep learning neural networks technique to detect rice diseases from the images. We developed an object detection model training and refinement process to improve the performance of our previous research on rice leave diseases detection. The process was based on analyzing the model's predictive results and could be repeatedly used to improve the quality of the database in the next training of the model. The deployment model for our LINE Bot system was created from the selected best performance technique in our previous paper, YOLOv3, trained by refined training data set. The performance of the deployment model was measured on 5 target classes and found that the Average True Positive Point improved from 91.1% in the previous paper to 95.6% in this study. Therefore, we used this deployment model for Rice Disease LINE Bot system. Our system worked automatically real-time to suggest primary diagnosis results to the users in the LINE group, which included rice farmers and rice disease specialists. They could communicate freely via chat. In the real LINE Bot deployment, the model's performance was measured by our own defined measurement Average True Positive Point and was found to be an average of 78.86%. The system was fast and took only 2-3 s for detection process in our system server.

(2 min) Designed to learn long-range interactions on sequential data, transformers continue to show state-of-the-art results on a wide variety of tasks. In contrast to CNNs, they contain no inductive bias that prioritizes local interactions. This makes them expressive, but also computationally infeasible for long sequences, such as high-resolution images. We demonstrate how combining the effectiveness of the inductive bias of CNNs with the expressivity of transformers enables them to model and thereby synthesize high-resolution images. We show how to (i) use CNNs to learn a context-rich vocabulary of image constituents, and in turn (ii) utilize transformers to efficiently model their composition within high-resolution images. Our approach is readily applied to conditional synthesis tasks, where both non-spatial information, such as object classes, and spatial information, such as segmentations, can control the generated image. In particular, we present the first results on semantically-guided synthesis of megapixel images with transformers and obtain the state of the art among autoregressive models on class-conditional ImageNet. Code and pretrained models can be found at https://github.com/CompVis/taming-transformers .

(2 min) Building instance segmentation models that are data-efficient and can handle rare object categories is an important challenge in computer vision. Leveraging data augmentations is a promising direction towards addressing this challenge. Here, we perform a systematic study of the Copy-Paste augmentation ([13, 12]) for instance segmentation where we randomly paste objects onto an image. Prior studies on Copy-Paste relied on modeling the surrounding visual context for pasting the objects. However, we find that the simple mechanism of pasting objects randomly is good enough and can provide solid gains on top of strong baselines. Furthermore, we show Copy-Paste is additive with semi-supervised methods that leverage extra data through pseudo labeling (e.g. self-training). On COCO instance segmentation, we achieve 49.1 mask AP and 57.3 box AP, an improvement of +0.6 mask AP and +1.5 box AP over the previous state-of-the-art. We further demonstrate that Copy-Paste can lead to significant improvements on the LVIS benchmark. Our baseline model outperforms the LVIS 2020 Challenge winning entry by +3.6 mask AP on rare categories.

(2 min) The outbreak of COVID-19 Disease due to the novel coronavirus has caused a shortage of medical resources. To aid and accelerate the diagnosis process, automatic diagnosis of COVID-19 via deep learning models has recently been explored by researchers across the world. While different data-driven deep learning models have been developed to mitigate the diagnosis of COVID-19, the data itself is still scarce due to patient privacy concerns. Federated Learning (FL) is a natural solution because it allows different organizations to cooperatively learn an effective deep learning model without sharing raw data. However, recent studies show that FL still lacks privacy protection and may cause data leakage. We investigate this challenging problem by proposing a simple yet effective algorithm, named \textbf{F}ederated \textbf{L}earning \textbf{o}n Medical Datasets using \textbf{P}artial Networks (FLOP), that shares only a partial model between the server and clients. Extensive experiments on benchmark data and real-world healthcare tasks show that our approach achieves comparable or better performance while reducing the privacy and security risks. Of particular interest, we conduct experiments on the COVID-19 dataset and find that our FLOP algorithm can allow different hospitals to collaboratively and effectively train a partially shared model without sharing local patients' data.

(3 min) Fluorescence microscopy allows for a detailed inspection of cells, cellular networks, and anatomical landmarks by staining with a variety of carefully-selected markers visualized as color channels. Quantitative characterization of structures in acquired images often relies on automatic image analysis methods. Despite the success of deep learning methods in other vision applications, their potential for fluorescence image analysis remains underexploited. One reason lies in the considerable workload required to train accurate models, which are normally specific for a given combination of markers, and therefore applicable to a very restricted number of experimental settings. We herein propose Marker Sampling and Excite, a neural network approach with a modality sampling strategy and a novel attention module that together enable (i) flexible training with heterogeneous datasets with combinations of markers and (ii) successful utility of learned models on arbitrary subsets of markers prospectively. We show that our single neural network solution performs comparably to an upper bound scenario where an ensemble of many networks is na\"ively trained for each possible marker combination separately. In addition, we demonstrate the feasibility of this framework in high-throughput biological analysis by revising a recent quantitative characterization of bone marrow vasculature in 3D confocal microscopy datasets and further confirm the validity of our approach on an additional, significantly different dataset of microvessels in fetal liver tissues. Not only can our work substantially ameliorate the use of deep learning in fluorescence microscopy analysis, but it can also be utilized in other fields with incomplete data acquisitions and missing modalities.

(2 min) Passive biometric identification enables wildlife monitoring with minimal disturbance. Using a motion-activated camera placed at an elevated position and facing downwards, we collected images of sea turtle carapace, each belonging to one of sixteen Chelonia mydas juveniles. We then learned co-variant and robust image descriptors from these images, enabling indexing and retrieval. In this work, we presented several classification results of sea turtle carapaces using the learned image descriptors. We found that a template-based descriptor, i.e., Histogram of Oriented Gradients (HOG) performed exceedingly better during classification than keypoint-based descriptors. For our dataset, a high-dimensional descriptor is a must due to the minimal gradient and color information inside the carapace images. Using HOG, we obtained an average classification accuracy of 65%.

(2 min) Cross-domain object detection is challenging, because object detection model is often vulnerable to data variance, especially to the considerable domain shift between two distinctive domains. In this paper, we propose a new Unbiased Mean Teacher (UMT) model for cross-domain object detection. We reveal that there often exists a considerable model bias for the simple mean teacher (MT) model in cross-domain scenarios, and eliminate the model bias with several simple yet highly effective strategies. In particular, for the teacher model, we propose a cross-domain distillation method for MT to maximally exploit the expertise of the teacher model. Moreover, for the student model, we alleviate its bias by augmenting training samples with pixel-level adaptation. Finally, for the teaching process, we employ an out-of-distribution estimation strategy to select samples that most fit the current model to further enhance the cross-domain distillation process. By tackling the model bias issue with these strategies, our UMT model achieves mAPs of 44.1%, 58.1%, 41.7%, and 43.1% on benchmark datasets Clipart1k, Watercolor2k, Foggy Cityscapes, and Cityscapes, respectively, which outperforms the existing state-of-the-art results in notable margins. Our implementation is available at https://github.com/kinredon/umt.

(3 min) The goal of this work is to characterize the representational impact that foveation operations have for machine vision systems, inspired by the foveated human visual system, which has higher acuity at the center of gaze and texture-like encoding in the periphery. To do so, we introduce models consisting of a first-stage \textit{fixed} image transform followed by a second-stage \textit{learnable} convolutional neural network, and we varied the first stage component. The primary model has a foveated-textural input stage, which we compare to a model with foveated-blurred input and a model with spatially-uniform blurred input (both matched for perceptual compression), and a final reference model with minimal input-based compression. We find that: 1) the foveated-texture model shows similar scene classification accuracy as the reference model despite its compressed input, with greater i.i.d. generalization than the other models; 2) the foveated-texture model has greater sensitivity to high-spatial frequency information and greater robustness to occlusion, w.r.t the comparison models; 3) both the foveated systems, show a stronger center image-bias relative to the spatially-uniform systems even with a weight sharing constraint. Critically, these results are preserved over different classical CNN architectures throughout their learning dynamics. Altogether, this suggests that foveation with peripheral texture-based computations yields an efficient, distinct, and robust representational format of scene information, and provides symbiotic computational insight into the representational consequences that texture-based peripheral encoding may have for processing in the human visual system, while also potentially inspiring the next generation of computer vision models via spatially-adaptive computation. Code + Data available here: https://github.com/ArturoDeza/EmergentProperties

(2 min) Unsupervised segmentation of action segments in egocentric videos is a desirable feature in tasks such as activity recognition and content-based video retrieval. Reducing the search space into a finite set of action segments facilitates a faster and less noisy matching. However, there exist a substantial gap in machine understanding of natural temporal cuts during a continuous human activity. This work reports on a novel gaze-based approach for segmenting action segments in videos captured using an egocentric camera. Gaze is used to locate the region-of-interest inside a frame. By tracking two simple motion-based parameters inside successive regions-of-interest, we discover a finite set of temporal cuts. We present several results using combinations (of the two parameters) on a dataset, i.e., BRISGAZE-ACTIONS. The dataset contains egocentric videos depicting several daily-living activities. The quality of the temporal cuts is further improved by implementing two entropy measures.

(2 min) We present a method to estimate an HDR environment map from a narrow field-of-view LDR camera image in real-time. This enables perceptually appealing reflections and shading on virtual objects of any material finish, from mirror to diffuse, rendered into a real physical environment using augmented reality. Our method is based on our efficient convolutional neural network architecture, EnvMapNet, trained end-to-end with two novel losses, ProjectionLoss for the generated image, and ClusterLoss for adversarial training. Through qualitative and quantitative comparison to state-of-the-art methods, we demonstrate that our algorithm reduces the directional error of estimated light sources by more than 50%, and achieves 3.7 times lower Frechet Inception Distance (FID). We further showcase a mobile application that is able to run our neural network model in under 9 ms on an iPhone XS, and render in real-time, visually coherent virtual objects in previously unseen real-world environments.

(2 min) Classification of rocks is one of the fundamental tasks in a geological study. The process requires a human expert to examine sampled thin section images under a microscope. In this study, we propose a method that uses microscope automation, digital image acquisition, edge detection and colour analysis (histogram). We collected 60 digital images from 20 standard thin sections using a digital camera mounted on a conventional microscope. Each image is partitioned into a finite number of cells that form a grid structure. Edge and colour profile of pixels inside each cell determine its classification. The individual cells then determine the thin section image classification via a majority voting scheme. Our method yielded successful results as high as 90% to 100% precision.

(2 min) Haustral folds are colon wall protrusions implicated for high polyp miss rate during optical colonoscopy procedures. If segmented accurately, haustral folds can allow for better estimation of missed surface and can also serve as valuable landmarks for registering pre-treatment virtual (CT) and optical colonoscopies, to guide navigation towards the anomalies found in pre-treatment scans. We present a novel generative adversarial network, FoldIt, for feature-consistent image translation of optical colonoscopy videos to virtual colonoscopy renderings with haustral fold overlays. A new transitive loss is introduced in order to leverage ground truth information between haustral fold annotations and virtual colonoscopy renderings. We demonstrate the effectiveness of our model on real challenging optical colonoscopy videos as well as on textured virtual colonoscopy videos with clinician-verified haustral fold annotations. All code and scripts to reproduce the experiments of this paper will be made available via our Computational Endoscopy Platform at https://github.com/nadeemlab/CEP.

(2 min) Left ventricular hypertrophy (LVH) results from chronic remodeling caused by a broad range of systemic and cardiovascular disease including hypertension, aortic stenosis, hypertrophic cardiomyopathy, and cardiac amyloidosis. Early detection and characterization of LVH can significantly impact patient care but is limited by under-recognition of hypertrophy, measurement error and variability, and difficulty differentiating etiologies of LVH. To overcome this challenge, we present EchoNet-LVH - a deep learning workflow that automatically quantifies ventricular hypertrophy with precision equal to human experts and predicts etiology of LVH. Trained on 28,201 echocardiogram videos, our model accurately measures intraventricular wall thickness (mean absolute error [MAE] 1.4mm, 95% CI 1.2-1.5mm), left ventricular diameter (MAE 2.4mm, 95% CI 2.2-2.6mm), and posterior wall thickness (MAE 1.2mm, 95% CI 1.1-1.3mm) and classifies cardiac amyloidosis (area under the curve of 0.83) and hypertrophic cardiomyopathy (AUC 0.98) from other etiologies of LVH. In external datasets from independent domestic and international healthcare systems, EchoNet-LVH accurately quantified ventricular parameters (R2 of 0.96 and 0.90 respectively) and detected cardiac amyloidosis (AUC 0.79) and hypertrophic cardiomyopathy (AUC 0.89) on the domestic external validation site. Leveraging measurements across multiple heart beats, our model can more accurately identify subtle changes in LV geometry and its causal etiologies. Compared to human experts, EchoNet-LVH is fully automated, allowing for reproducible, precise measurements, and lays the foundation for precision diagnosis of cardiac hypertrophy. As a resource to promote further innovation, we also make publicly available a large dataset of 23,212 annotated echocardiogram videos.

(2 min) The subtleties of human perception, as measured by vision scientists through the use of psychophysics, are important clues to the internal workings of visual recognition. For instance, measured reaction time can indicate whether a visual stimulus is easy for a subject to recognize, or whether it is hard. In this paper, we consider how to incorporate psychophysical measurements of visual perception into the loss function of a deep neural network being trained for a recognition task, under the assumption that such information can enforce consistency with human behavior. As a case study to assess the viability of this approach, we look at the problem of handwritten document transcription. While good progress has been made towards automatically transcribing modern handwriting, significant challenges remain in transcribing historical documents. Here we describe a general enhancement strategy, underpinned by the new loss formulation, which can be applied to the training regime of any deep learning-based document transcription system. Through experimentation, reliable performance improvement is demonstrated for the standard IAM and RIMES datasets for three different network architectures. Further, we go on to show feasibility for our approach on a new dataset of digitized Latin manuscripts, originally produced by scribes in the Cloister of St. Gall in the the 9th century.

(2 min) Multimodal VAEs seek to model the joint distribution over heterogeneous data (e.g.\ vision, language), whilst also capturing a shared representation across such modalities. Prior work has typically combined information from the modalities by reconciling idiosyncratic representations directly in the recognition model through explicit products, mixtures, or other such factorisations. Here we introduce a novel alternative, the MEME, that avoids such explicit combinations by repurposing semi-supervised VAEs to combine information between modalities implicitly through mutual supervision. This formulation naturally allows learning from partially-observed data where some modalities can be entirely missing -- something that most existing approaches either cannot handle, or do so to a limited extent. We demonstrate that MEME outperforms baselines on standard metrics across both partial and complete observation schemes on the MNIST-SVHN (image-image) and CUB (image-text) datasets. We also contrast the quality of the representations learnt by mutual supervision against standard approaches and observe interesting trends in its ability to capture relatedness between data.

(2 min) Accurate detection of obstacles in 3D is an essential task for autonomous driving and intelligent transportation. In this work, we propose a general multimodal fusion framework FusionPainting to fuse the 2D RGB image and 3D point clouds at a semantic level for boosting the 3D object detection task. Especially, the FusionPainting framework consists of three main modules: a multi-modal semantic segmentation module, an adaptive attention-based semantic fusion module, and a 3D object detector. First, semantic information is obtained for 2D images and 3D Lidar point clouds based on 2D and 3D segmentation approaches. Then the segmentation results from different sensors are adaptively fused based on the proposed attention-based semantic fusion module. Finally, the point clouds painted with the fused semantic label are sent to the 3D detector for obtaining the 3D objection results. The effectiveness of the proposed framework has been verified on the large-scale nuScenes detection benchmark by comparing it with three different baselines. The experimental results show that the fusion strategy can significantly improve the detection performance compared to the methods using only point clouds, and the methods using point clouds only painted with 2D segmentation information. Furthermore, the proposed approach outperforms other state-of-the-art methods on the nuScenes testing benchmark.

(2 min) Unsupervised domain adaptation (UDA) aims to transfer knowledge learned from a labeled source domain to an unlabeled and unseen target domain, which is usually trained on data from both domains. Access to the source domain data at the adaptation stage, however, is often limited, due to data storage or privacy issues. To alleviate this, in this work, we target source free UDA for segmentation, and propose to adapt an ``off-the-shelf" segmentation model pre-trained in the source domain to the target domain, with an adaptive batch-wise normalization statistics adaptation framework. Specifically, the domain-specific low-order batch statistics, i.e., mean and variance, are gradually adapted with an exponential momentum decay scheme, while the consistency of domain shareable high-order batch statistics, i.e., scaling and shifting parameters, is explicitly enforced by our optimization objective. The transferability of each channel is adaptively measured first from which to balance the contribution of each channel. Moreover, the proposed source free UDA framework is orthogonal to unsupervised learning methods, e.g., self-entropy minimization, which can thus be simply added on top of our framework. Extensive experiments on the BraTS 2018 database show that our source free UDA framework outperformed existing source-relaxed UDA methods for the cross-subtype UDA segmentation task and yielded comparable results for the cross-modality UDA segmentation task, compared with a supervised UDA methods with the source data.

(2 min) Diabetic Retinopathy (DR) is among the worlds leading vision loss causes in diabetic patients. DR is a microvascular disease that affects the eye retina, which causes vessel blockage and therefore cuts the main source of nutrition for the retina tissues. Treatment for this visual disorder is most effective when it is detected in its earliest stages, as severe DR can result in irreversible blindness. Nonetheless, DR identification requires the expertise of Ophthalmologists which is often expensive and time-consuming. Therefore, automatic detection systems were introduced aiming to facilitate the identification process, making it available globally in a time and cost-efficient manner. However, due to the limited reliable datasets and medical records for this particular eye disease, the obtained predictions accuracies were relatively unsatisfying for eye specialists to rely on them as diagnostic systems. Thus, we explored an ensemble-based learning strategy, merging a substantial selection of well-known classification algorithms in one sophisticated diagnostic model. The proposed framework achieved the highest accuracy rates among all other common classification algorithms in the area. 4 subdatasets were generated to contain the top 5 and top 10 features of the Messidor dataset, selected by InfoGainEval. and WrapperSubsetEval., accuracies of 70.7% and 75.1% were achieved on the InfoGainEval. top 5 and original dataset respectively. The results imply the impressive performance of the subdataset, which significantly conduces to a less complex classification process

(2 min) The diagnosis of blood-based diseases often involves identifying and characterizing patient blood samples. Automated methods to detect and classify blood cell subtypes have important medical applications. Automated medical image processing and analysis offers a powerful tool for medical diagnosis. In this work we tackle the problem of white blood cell classification based on the morphological characteristics of their outer contour, color. The work we would explore a set of preprocessing and segmentation (Color-based segmentation, Morphological processing, contouring) algorithms along with a set of features extraction methods (Corner detection algorithms and Histogram of Gradients(HOG)), dimensionality reduction algorithms (Principal Component Analysis(PCA)) that are able to recognize and classify through various Unsupervised(k-nearest neighbors) and Supervised (Support Vector Machine, Decision Trees, Linear Discriminant Analysis, Quadratic Discriminant Analysis, Naive Bayes) algorithms different categories of white blood cells to Eosinophil, Lymphocyte, Monocyte, and Neutrophil. We even take a step forwards to explore various Deep Convolutional Neural network architecture (Sqeezent, MobilenetV1,MobilenetV2, InceptionNet etc.) without preprocessing/segmentation and with preprocessing. We would like to explore many algorithms to identify the robust algorithm with least time complexity and low resource requirement. The outcome of this work can be a cue to selection of algorithms as per requirement for automated blood cell classification.

(3 min) The subject of "fairness" in artificial intelligence (AI) refers to assessing AI algorithms for potential bias based on demographic characteristics such as race and gender, and the development of algorithms to address this bias. Most applications to date have been in computer vision, although some work in healthcare has started to emerge. The use of deep learning (DL) in cardiac MR segmentation has led to impressive results in recent years, and such techniques are starting to be translated into clinical practice. However, no work has yet investigated the fairness of such models. In this work, we perform such an analysis for racial/gender groups, focusing on the problem of training data imbalance, using a nnU-Net model trained and evaluated on cine short axis cardiac MR data from the UK Biobank dataset, consisting of 5,903 subjects from 6 different racial groups. We find statistically significant differences in Dice performance between different racial groups. To reduce the racial bias, we investigated three strategies: (1) stratified batch sampling, in which batch sampling is stratified to ensure balance between racial groups; (2) fair meta-learning for segmentation, in which a DL classifier is trained to classify race and jointly optimized with the segmentation model; and (3) protected group models, in which a different segmentation model is trained for each racial group. We also compared the results to the scenario where we have a perfectly balanced database. To assess fairness we used the standard deviation (SD) and skewed error ratio (SER) of the average Dice values. Our results demonstrate that the racial bias results from the use of imbalanced training data, and that all proposed bias mitigation strategies improved fairness, with the best SD and SER resulting from the use of protected group models.

(2 min) The coronavirus continues to disrupt our everyday lives as it spreads at an exponential rate. It needs to be detected quickly in order to quarantine positive patients so as to avoid further spread. This work proposes a new convolutional neural network (CNN) architecture called 'slow Encoding CNN. The proposed model's best performance wrt Sensitivity, Positive Predictive Value (PPV) found to be SP=0.67, PP=0.98, SN=0.96, and PN=0.52 on AI AGAINST COVID19 - Screening X-ray images for COVID-19 Infections competition's test data samples. SP and PP stand for the Sensitivity and PPV of the COVID-19 positive class, while PN and SN stand for the Sensitivity and PPV of the COVID-19 negative class.

(2 min) 3D face reconstruction from a single image is a task that has garnered increased interest in the Computer Vision community, especially due to its broad use in a number of applications such as realistic 3D avatar creation, pose invariant face recognition and face hallucination. Since the introduction of the 3D Morphable Model in the late 90's, we witnessed an explosion of research aiming at particularly tackling this task. Nevertheless, despite the increasing level of detail in the 3D face reconstructions from single images mainly attributed to deep learning advances, finer and highly deformable components of the face such as the tongue are still absent from all 3D face models in the literature, although being very important for the realness of the 3D avatar representations. In this work we present the first, to the best of our knowledge, end-to-end trainable pipeline that accurately reconstructs the 3D face together with the tongue. Moreover, we make this pipeline robust in "in-the-wild" images by introducing a novel GAN method tailored for 3D tongue surface generation. Finally, we make publicly available to the community the first diverse tongue dataset, consisting of 1,800 raw scans of 700 individuals varying in gender, age, and ethnicity backgrounds. As we demonstrate in an extensive series of quantitative as well as qualitative experiments, our model proves to be robust and realistically captures the 3D tongue structure, even in adverse "in-the-wild" conditions.

(2 min) Low-rankness is important in the hyperspectral image (HSI) denoising tasks. The tensor nuclear norm (TNN), defined based on the tensor singular value decomposition, is a state-of-the-art method to describe the low-rankness of HSI. However, TNN ignores some of the physical meanings of HSI in tackling the denoising tasks, leading to suboptimal denoising performance. In this paper, we propose the multi-modal and frequency-weighted tensor nuclear norm (MFWTNN) and the non-convex MFWTNN for HSI denoising tasks. Firstly, we investigate the physical meaning of frequency components and reconsider their weights to improve the low-rank representation ability of TNN. Meanwhile, we also consider the correlation among two spatial dimensions and the spectral dimension of HSI and combine the above improvements to TNN to propose MFWTNN. Secondly, we use non-convex functions to approximate the rank function of the frequency tensor and propose the NonMFWTNN to relax the MFWTNN better. Besides, we adaptively choose bigger weights for slices mainly containing noise information and smaller weights for slices containing profile information. Finally, we develop the efficient alternating direction method of multiplier (ADMM) based algorithm to solve the proposed models, and the effectiveness of our models are substantiated in simulated and real HSI datasets.

(2 min) Visual Emotion Analysis (VEA) has attracted increasing attention recently with the prevalence of sharing images on social networks. Since human emotions are ambiguous and subjective, it is more reasonable to address VEA in a label distribution learning (LDL) paradigm rather than a single-label classification task. Different from other LDL tasks, there exist intrinsic relationships between emotions and unique characteristics within them, as demonstrated in psychological theories. Inspired by this, we propose a well-grounded circular-structured representation to utilize the prior knowledge for visual emotion distribution learning. To be specific, we first construct an Emotion Circle to unify any emotional state within it. On the proposed Emotion Circle, each emotion distribution is represented with an emotion vector, which is defined with three attributes (i.e., emotion polarity, emotion type, emotion intensity) as well as two properties (i.e., similarity, additivity). Besides, we design a novel Progressive Circular (PC) loss to penalize the dissimilarities between predicted emotion vector and labeled one in a coarse-to-fine manner, which further boosts the learning process in an emotion-specific way. Extensive experiments and comparisons are conducted on public visual emotion distribution datasets, and the results demonstrate that the proposed method outperforms the state-of-the-art methods.

(2 min) Binarized Neural Networks (BNNs) have the potential to revolutionize the way that deep learning is carried out in edge computing platforms. However, the effectiveness of interpretability methods on these networks has not been assessed. In this paper, we compare the performance of several widely used saliency map-based interpretabilty techniques (Gradient, SmoothGrad and GradCAM), when applied to Binarized or Full Precision Neural Networks (FPNNs). We found that the basic Gradient method produces very similar-looking maps for both types of network. However, SmoothGrad produces significantly noisier maps for BNNs. GradCAM also produces saliency maps which differ between network types, with some of the BNNs having seemingly nonsensical explanations. We comment on possible reasons for these differences in explanations and present it as an example of why interpretability techniques should be tested on a wider range of network types.

(2 min) Sketch drawings capture the salient information of visual concepts. Previous work has shown that neural networks are capable of producing sketches of natural objects drawn from a small number of classes. While earlier approaches focus on generation quality or retrieval, we explore properties of image representations learned by training a model to produce sketches of images. We show that this generative, class-agnostic model produces informative embeddings of images from novel examples, classes, and even novel datasets in a few-shot setting. Additionally, we find that these learned representations exhibit interesting structure and compositionality.

(2 min) We introduce feature alignment, a technique for obtaining approximate reversibility in artificial neural networks. By means of feature extraction, we can train a neural network to learn an estimated map for its reverse process from outputs to inputs. Combined with variational autoencoders, we can generate new samples from the same statistics as the training data. Improvements of the results are obtained by using concepts from generative adversarial networks. Finally, we show that the technique can be modified for training neural networks locally, saving computational memory resources. Applying these techniques, we report results for three vision generative tasks: MNIST, CIFAR-10, and celebA.

(2 min) Semantic segmentation from very fine resolution (VFR) urban scene images plays a significant role in several application scenarios including autonomous driving, land cover classification, and urban planning, etc. However, the tremendous details contained in the VFR image severely limit the potential of the existing deep learning approaches. More seriously, the considerable variations in scale and appearance of objects further deteriorate the representational capacity of those se-mantic segmentation methods, leading to the confusion of adjacent objects. Addressing such is-sues represents a promising research field in the remote sensing community, which paves the way for scene-level landscape pattern analysis and decision making. In this manuscript, we pro-pose a bilateral awareness network (BANet) which contains a dependency path and a texture path to fully capture the long-range relationships and fine-grained details in VFR images. Specif-ically, the dependency path is conducted based on the ResT, a novel Transformer backbone with memory-efficient multi-head self-attention, while the texture path is built on the stacked convo-lution operation. Besides, using the linear attention mechanism, a feature aggregation module (FAM) is designed to effectively fuse the dependency features and texture features. Extensive experiments conducted on the three large-scale urban scene image segmentation datasets, i.e., ISPRS Vaihingen dataset, ISPRS Potsdam dataset, and UAVid dataset, demonstrate the effective-ness of our BANet. Specifically, a 64.6% mIoU is achieved on the UAVid dataset.

(2 min) Recent studies have shown remarkable success in the unsupervised image to image (I2I) translation. However, due to the imbalance in the data, learning joint distribution for various domains is still very challenging. Although existing models can generate realistic target images, it's difficult to maintain the structure of the source image. In addition, training a generative model on large data in multiple domains requires a lot of time and computer resources. To address these limitations, we propose a novel image-to-image translation method that generates images of the target domain by finetuning a stylegan2 pretrained model. The stylegan2 model is suitable for unsupervised I2I translation on unbalanced datasets; it is highly stable, produces realistic images, and even learns properly from limited data when applied with simple fine-tuning techniques. Thus, in this paper, we propose new methods to preserve the structure of the source images and generate realistic images in the target domain. The code and results are available at https://github.com/happy-jihye/Cartoon-StyleGan2

(2 min) Accurate prediction of pedestrian and bicyclist paths is integral to the development of reliable autonomous vehicles in dense urban environments. The interactions between vehicle and pedestrian or bicyclist have a significant impact on the trajectories of traffic participants e.g. stopping or turning to avoid collisions. Although recent datasets and trajectory prediction approaches have fostered the development of autonomous vehicles yet the amount of vehicle-pedestrian (bicyclist) interactions modeled are sparse. In this work, we propose Euro-PVI, a dataset of pedestrian and bicyclist trajectories. In particular, our dataset caters more diverse and complex interactions in dense urban scenarios compared to the existing datasets. To address the challenges in predicting future trajectories with dense interactions, we develop a joint inference model that learns an expressive multi-modal shared latent space across agents in the urban scene. This enables our Joint-$\beta$-cVAE approach to better model the distribution of future trajectories. We achieve state of the art results on the nuScenes and Euro-PVI datasets demonstrating the importance of capturing interactions between ego-vehicle and pedestrians (bicyclists) for accurate predictions.

(2 min) We observe that despite their hierarchical convolutional nature, the synthesis process of typical generative adversarial networks depends on absolute pixel coordinates in an unhealthy manner. This manifests itself as, e.g., detail appearing to be glued to image coordinates instead of the surfaces of depicted objects. We trace the root cause to careless signal processing that causes aliasing in the generator network. Interpreting all signals in the network as continuous, we derive generally applicable, small architectural changes that guarantee that unwanted information cannot leak into the hierarchical synthesis process. The resulting networks match the FID of StyleGAN2 but differ dramatically in their internal representations, and they are fully equivariant to translation and rotation even at subpixel scales. Our results pave the way for generative models better suited for video and animation.

(2 min) In this paper we propose the first deep unsupervised approach in human body reconstruction to estimate body surface from a sparse set of landmarks, so called DeepMurf. We apply a denoising autoencoder to estimate missing landmarks. Then we apply an attention model to estimate body joints from landmarks. Finally, a cascading network is applied to regress parameters of a statistical generative model that reconstructs body. Our set of proposed loss functions allows us to train the network in an unsupervised way. Results on four public datasets show that our approach accurately reconstructs the human body from real world mocap data.

(2 min) The natural world is long-tailed: rare classes are observed orders of magnitudes less frequently than common ones, leading to highly-imbalanced data where rare classes can have only handfuls of examples. Learning from few examples is a known challenge for deep learning based classification algorithms, and is the focus of the field of low-shot learning. One potential approach to increase the training data for these rare classes is to augment the limited real data with synthetic samples. This has been shown to help, but the domain shift between real and synthetic hinders the approaches' efficacy when tested on real data. We explore the use of image-to-image translation methods to close the domain gap between synthetic and real imagery for animal species classification in data collected from camera traps: motion-activated static cameras used to monitor wildlife. We use low-level feature alignment between source and target domains to make synthetic data for a rare species generated using a graphics engine more "realistic". Compared against a system augmented with unaligned synthetic data, our experiments show a considerable decrease in classification error rates on a rare species.

(2 min) Diabetic retinopathy (DR) remains the most prevalent cause of vision impairment and irreversible blindness in the working-age adults. Due to the renaissance of deep learning (DL), DL-based DR diagnosis has become a promising tool for the early screening and severity grading of DR. However, training deep neural networks (DNNs) requires an enormous amount of carefully labeled data. Noisy label data may be introduced when labeling plenty of data, degrading the performance of models. In this work, we propose a novel label management mechanism (LMM) for the DNN to overcome overfitting on the noisy data. LMM utilizes maximum posteriori probability (MAP) in the Bayesian statistic and time-weighted technique to selectively correct the labels of unclean data, which gradually purify the training data and improve classification performance. Comprehensive experiments on both synthetic noise data (Messidor \& our collected DR dataset) and real-world noise data (ANIMAL-10N) demonstrated that LMM could boost performance of models and is superior to three state-of-the-art methods.

(2 min) Background noise and scale variation are common problems that have been long recognized in crowd counting. Humans glance at a crowd image and instantly know the approximate number of human and where they are through attention the crowd regions and the congestion degree of crowd regions with a global receptive filed. Hence, in this paper, we propose a novel feedback network with Region-Aware block called RANet by modeling human's Top-Down visual perception mechanism. Firstly, we introduce a feedback architecture to generate priority maps that provide prior about candidate crowd regions in input images. The prior enables the RANet pay more attention to crowd regions. Then we design Region-Aware block that could adaptively encode the contextual information into input images through global receptive field. More specifically, we scan the whole input images and its priority maps in the form of column vector to obtain a relevance matrix estimating their similarity. The relevance matrix obtained would be utilized to build global relationships between pixels. Our method outperforms state-of-the-art crowd counting methods on several public datasets.

(2 min) One widely used approach towards understanding the inner workings of deep convolutional neural networks is to visualize unit responses via activation maximization. Feature visualizations via activation maximization are thought to provide humans with precise information about the image features that cause a unit to be activated. If this is indeed true, these synthetic images should enable humans to predict the effect of an intervention, such as whether occluding a certain patch of the image (say, a dog's head) changes a unit's activation. Here, we test this hypothesis by asking humans to predict which of two square occlusions causes a larger change to a unit's activation. Both a large-scale crowdsourced experiment and measurements with experts show that on average, the extremely activating feature visualizations by Olah et al. (2017) indeed help humans on this task ($67 \pm 4\%$ accuracy; baseline performance without any visualizations is $60 \pm 3\%$). However, they do not provide any significant advantage over other visualizations (such as e.g. dataset samples), which yield similar performance ($66 \pm 3\%$ to $67 \pm 3\%$ accuracy). Taken together, we propose an objective psychophysical task to quantify the benefit of unit-level interpretability methods for humans, and find no evidence that feature visualizations provide humans with better "causal understanding" than simple alternative visualizations.

(2 min) In this paper, we present Vision Permutator, a conceptually simple and data efficient MLP-like architecture for visual recognition. By realizing the importance of the positional information carried by 2D feature representations, unlike recent MLP-like models that encode the spatial information along the flattened spatial dimensions, Vision Permutator separately encodes the feature representations along the height and width dimensions with linear projections. This allows Vision Permutator to capture long-range dependencies along one spatial direction and meanwhile preserve precise positional information along the other direction. The resulting position-sensitive outputs are then aggregated in a mutually complementing manner to form expressive representations of the objects of interest. We show that our Vision Permutators are formidable competitors to convolutional neural networks (CNNs) and vision transformers. Without the dependence on spatial convolutions or attention mechanisms, Vision Permutator achieves 81.5% top-1 accuracy on ImageNet without extra large-scale training data (e.g., ImageNet-22k) using only 25M learnable parameters, which is much better than most CNNs and vision transformers under the same model size constraint. When scaling up to 88M, it attains 83.2% top-1 accuracy. We hope this work could encourage research on rethinking the way of encoding spatial information and facilitate the development of MLP-like models. Code is available at https://github.com/Andrew-Qibin/VisionPermutator.

(2 min) A large scale human-labeled dataset plays an important role in creating high quality deep learning models. In this paper we present text annotation for Open Images V5 dataset. To our knowledge it is the largest among publicly available manually created text annotations. Having this annotation we trained a simple Mask-RCNN-based network, referred as Yet Another Mask Text Spotter (YAMTS), which achieves competitive performance or even outperforms current state-of-the-art approaches in some cases on ICDAR2013, ICDAR2015 and Total-Text datasets. Code for text spotting model available online at: https://github.com/openvinotoolkit/training_extensions. The model can be exported to OpenVINO-format and run on Intel CPUs.

(2 min) The abundance of clouds, located both spatially and temporally, often makes remote sensing applications with optical images difficult or even impossible. In this manuscript, a novel method for clouds-corrupted optical image restoration has been presented and developed, based on a joint data fusion paradigm, where three deep neural networks have been combined in order to fuse spatio-temporal features extracted from Sentinel-1 and Sentinel-2 time-series of data. It is worth highlighting that both the code and the dataset have been implemented from scratch and made available to interested research for further analysis and investigation.

(2 min) Transformers recently are adapted from the community of natural language processing as a promising substitute of convolution-based neural networks for visual learning tasks. However, its supremacy degenerates given an insufficient amount of training data (e.g., ImageNet). To make it into practical utility, we propose a novel distillation-based method to train vision transformers. Unlike previous works, where merely heavy convolution-based teachers are provided, we introduce lightweight teachers with different architectural inductive biases (e.g., convolution and involution) to co-advise the student transformer. The key is that teachers with different inductive biases attain different knowledge despite that they are trained on the same dataset, and such different knowledge compounds and boosts the student's performance during distillation. Equipped with this cross inductive bias distillation method, our vision transformers (termed as CivT) outperform all previous transformers of the same architecture on ImageNet.

(2 min) Over the last decade, the development of deep image classification networks has mostly been driven by the search for the best performance in terms of classification accuracy on standardized benchmarks like ImageNet. More recently, this focus has been expanded by the notion of model robustness, i.e. the generalization abilities of models towards previously unseen changes in the data distribution. While new benchmarks, like ImageNet-C, have been introduced to measure robustness properties, we argue that fixed testsets are only able to capture a small portion of possible data variations and are thus limited and prone to generate new overfitted solutions. To overcome these drawbacks, we suggest to estimate the robustness of a model directly from the structure of its learned feature-space. We introduce robustness indicators which are obtained via unsupervised clustering of latent representations inside a trained classifier and show very high correlations to the model performance on corrupted test data.

(2 min) Self-training based unsupervised domain adaptation (UDA) has shown great potential to address the problem of domain shift, when applying a trained deep learning model in a source domain to unlabeled target domains. However, while the self-training UDA has demonstrated its effectiveness on discriminative tasks, such as classification and segmentation, via the reliable pseudo-label selection based on the softmax discrete histogram, the self-training UDA for generative tasks, such as image synthesis, is not fully investigated. In this work, we propose a novel generative self-training (GST) UDA framework with continuous value prediction and regression objective for cross-domain image synthesis. Specifically, we propose to filter the pseudo-label with an uncertainty mask, and quantify the predictive confidence of generated images with practical variational Bayes learning. The fast test-time adaptation is achieved by a round-based alternative optimization scheme. We validated our framework on the tagged-to-cine magnetic resonance imaging (MRI) synthesis problem, where datasets in the source and target domains were acquired from different scanners or centers. Extensive validations were carried out to verify our framework against popular adversarial training UDA methods. Results show that our GST, with tagged MRI of test subjects in new target domains, improved the synthesis quality by a large margin, compared with the adversarial training UDA methods.

(2 min) The efficiency and accuracy of mapping are crucial in a large scene and long-term AR applications. Multi-agent cooperative SLAM is the precondition of multi-user AR interaction. The cooperation of multiple smart phones has the potential to improve efficiency and robustness of task completion and can complete tasks that a single agent cannot do. However, it depends on robust communication, efficient location detection, robust mapping, and efficient information sharing among agents. We propose a multi-intelligence collaborative monocular visual-inertial SLAM deployed on multiple ios mobile devices with a centralized architecture. Each agent can independently explore the environment, run a visual-inertial odometry module online, and then send all the measurement information to a central server with higher computing resources. The server manages all the information received, detects overlapping areas, merges and optimizes the map, and shares information with the agents when needed. We have verified the performance of the system in public datasets and real environments. The accuracy of mapping and fusion of the proposed system is comparable to VINS-Mono which requires higher computing resources.

(2 min) This work tackles the issue of fairness in the context of generative procedures, such as image super-resolution, which entail different definitions from the standard classification setting. Moreover, while traditional group fairness definitions are typically defined with respect to specified protected groups -- camouflaging the fact that these groupings are artificial and carry historical and political motivations -- we emphasize that there are no ground truth identities. For instance, should South and East Asians be viewed as a single group or separate groups? Should we consider one race as a whole or further split by gender? Choosing which groups are valid and who belongs in them is an impossible dilemma and being ``fair'' with respect to Asians may require being ``unfair'' with respect to South Asians. This motivates the introduction of definitions that allow algorithms to be \emph{oblivious} to the relevant groupings. We define several intuitive notions of group fairness and study their incompatibilities and trade-offs. We show that the natural extension of demographic parity is strongly dependent on the grouping, and \emph{impossible} to achieve obliviously. On the other hand, the conceptually new definition we introduce, Conditional Proportional Representation, can be achieved obliviously through Posterior Sampling. Our experiments validate our theoretical results and achieve fair image reconstruction using state-of-the-art generative models.

(2 min) Federated Learning (FL) has become an active and promising distributed machine learning paradigm. As a result of statistical heterogeneity, recent studies clearly show that the performance of popular FL methods (e.g., FedAvg) deteriorates dramatically due to the client drift caused by local updates. This paper proposes a novel Federated Learning algorithm (called IGFL), which leverages both Individual and Group behaviors to mimic distribution, thereby improving the ability to deal with heterogeneity. Unlike existing FL methods, our IGFL can be applied to both client and server optimization. As a by-product, we propose a new attention-based federated learning in the server optimization of IGFL. To the best of our knowledge, this is the first time to incorporate attention mechanisms into federated optimization. We conduct extensive experiments and show that IGFL can significantly improve the performance of existing federated learning methods. Especially when the distributions of data among individuals are diverse, IGFL can improve the classification accuracy by about 13% compared with prior baselines.

(2 min) The Coronavirus disease 2019 (COVID-19) has rapidly spread all over the world since its first report in December 2019 and thoracic computed tomography (CT) has become one of the main tools for its diagnosis. In recent years, deep learning-based approaches have shown impressive performance in myriad image recognition tasks. However, they usually require a large number of annotated data for training. Inspired by Ground Glass Opacity (GGO), a common finding in COIVD-19 patient's CT scans, we proposed in this paper a novel self-supervised pretraining method based on pseudo lesions generation and restoration for COVID-19 diagnosis. We used Perlin noise, a gradient noise based mathematical model, to generate lesion-like patterns, which were then randomly pasted to the lung regions of normal CT images to generate pseudo COVID-19 images. The pairs of normal and pseudo COVID-19 images were then used to train an encoder-decoder architecture based U-Net for image restoration, which does not require any labelled data. The pretrained encoder was then fine-tuned using labelled data for COVID-19 diagnosis task. Two public COVID-19 diagnosis datasets made up of CT images were employed for evaluation. Comprehensive experimental results demonstrated that the proposed self-supervised learning approach could extract better feature representation for COVID-19 diagnosis and the accuracy of the proposed method outperformed the supervised model pretrained on large scale images by 6.57% and 3.03% on SARS-CoV-2 dataset and Jinan COVID-19 dataset, respectively.

(2 min) Fetal motion is unpredictable and rapid on the scale of conventional MR scan times. Therefore, dynamic fetal MRI, which aims at capturing fetal motion and dynamics of fetal function, is limited to fast imaging techniques with compromises in image quality and resolution. Super-resolution for dynamic fetal MRI is still a challenge, especially when multi-oriented stacks of image slices for oversampling are not available and high temporal resolution for recording the dynamics of the fetus or placenta is desired. Further, fetal motion makes it difficult to acquire high-resolution images for supervised learning methods. To address this problem, in this work, we propose STRESS (Spatio-Temporal Resolution Enhancement with Simulated Scans), a self-supervised super-resolution framework for dynamic fetal MRI with interleaved slice acquisitions. Our proposed method simulates an interleaved slice acquisition along the high-resolution axis on the originally acquired data to generate pairs of low- and high-resolution images. Then, it trains a super-resolution network by exploiting both spatial and temporal correlations in the MR time series, which is used to enhance the resolution of the original data. Evaluations on both simulated and in utero data show that our proposed method outperforms other self-supervised super-resolution methods and improves image quality, which is beneficial to other downstream tasks and evaluations.

(2 min) We propose a new method for unsupervised continual knowledge consolidation in generative models that relies on the partitioning of Variational Autoencoder's latent space. Acquiring knowledge about new data samples without forgetting previous ones is a critical problem of continual learning. Currently proposed methods achieve this goal by extending the existing model while constraining its behavior not to degrade on the past data, which does not exploit the full potential of relations within the entire training dataset. In this work, we identify this limitation and posit the goal of continual learning as a knowledge accumulation task. We solve it by continuously re-aligning latent space partitions that we call bands which are representations of samples seen in different tasks, driven by the similarity of the information they contain. In addition, we introduce a simple yet effective method for controlled forgetting of past data that improves the quality of reconstructions encoded in latent bands and a latent space disentanglement technique that improves knowledge consolidation. On top of the standard continual learning evaluation benchmarks, we evaluate our method on a new knowledge consolidation scenario and show that the proposed approach outperforms state-of-the-art by up to twofold across all testing scenarios.

(2 min) Histological subtype of papillary (p) renal cell carcinoma (RCC), type 1 vs. type 2, is an essential prognostic factor. The two subtypes of pRCC have a similar pattern, i.e., the papillary architecture, yet some subtle differences, including cellular and cell-layer level patterns. However, the cellular and cell-layer level patterns almost cannot be captured by existing CNN-based models in large-size histopathological images, which brings obstacles to directly applying these models to such a fine-grained classification task. This paper proposes a novel instance-based Vision Transformer (i-ViT) to learn robust representations of histopathological images for the pRCC subtyping task by extracting finer features from instance patches (by cropping around segmented nuclei and assigning predicted grades). The proposed i-ViT takes top-K instances as input and aggregates them for capturing both the cellular and cell-layer level patterns by a position-embedding layer, a grade-embedding layer, and a multi-head multi-layer self-attention module. To evaluate the performance of the proposed framework, experienced pathologists are invited to selected 1162 regions of interest from 171 whole slide images of type 1 and type 2 pRCC. Experimental results show that the proposed method achieves better performance than existing CNN-based models with a significant margin.

(2 min) In today's world, the amount of data produced in every field has increased at an unexpected level. In the face of increasing data, the importance of data processing has increased remarkably. Our resource topic is on the processing of video data, which has an important place in increasing data, and the production of summary videos. Within the scope of this resource, a new method for anomaly detection with object-based unsupervised learning has been developed while creating a video summary. By using this method, the video data is processed as pixels and the result is produced as a video segment. The process flow can be briefly summarized as follows. Objects on the video are detected according to their type, and then they are tracked. Then, the tracking history data of the objects are processed, and the classifier is trained with the object type. Thanks to this classifier, anomaly behavior of objects is detected. Video segments are determined by processing video moments containing anomaly behaviors. The video summary is created by extracting the detected video segments from the original video and combining them. The model we developed has been tested and verified separately for single camera and dual camera systems.

(2 min) Image captioning has increasingly large domains of application, and fashion is not an exception. Having automatic item descriptions is of great interest for fashion web platforms hosting sometimes hundreds of thousands of images. This paper is one of the first tackling image captioning for fashion images. To contribute addressing dataset diversity issues, we introduced the InFashAIv1 dataset containing almost 16.000 African fashion item images with their titles, prices and general descriptions. We also used the well known DeepFashion dataset in addition to InFashAIv1. Captions are generated using the \textit{Show and Tell} model made of CNN encoder and RNN Decoder. We showed that jointly training the model on both datasets improves captions quality for African style fashion images, suggesting a transfer learning from Western style data. The InFashAIv1 dataset is released on \href{https://github.com/hgilles06/infashai}{Github} to encourage works with more diversity inclusion.

(2 min) Although instance segmentation has made considerable advancement over recent years, it's still a challenge to design high accuracy algorithms with real-time performance. In this paper, we propose a real-time instance segmentation framework termed OrienMask. Upon the one-stage object detector YOLOv3, a mask head is added to predict some discriminative orientation maps, which are explicitly defined as spatial offset vectors for both foreground and background pixels. Thanks to the discrimination ability of orientation maps, masks can be recovered without the need for extra foreground segmentation. All instances that match with the same anchor size share a common orientation map. This special sharing strategy reduces the amortized memory utilization for mask predictions but without loss of mask granularity. Given the surviving box predictions after NMS, instance masks can be concurrently constructed from the corresponding orientation maps with low complexity. Owing to the concise design for mask representation and its effective integration with the anchor-based object detector, our method is qualified under real-time conditions while maintaining competitive accuracy. Experiments on COCO benchmark show that OrienMask achieves 34.8 mask AP at the speed of 42.7 fps evaluated with a single RTX 2080 Ti. The code is available at https://github.com/duwt/OrienMask.

(2 min) We introduce Transductive Infomation Maximization (TIM) for few-shot learning. Our method maximizes the mutual information between the query features and their label predictions for a given few-shot task, in conjunction with a supervision loss based on the support set. We motivate our transductive loss by deriving a formal relation between the classification accuracy and mutual-information maximization. Furthermore, we propose a new alternating-direction solver, which substantially speeds up transductive inference over gradient-based optimization, while yielding competitive accuracy. We also provide a convergence analysis of our solver based on Zangwill's theory and bound-optimization arguments. TIM inference is modular: it can be used on top of any base-training feature extractor. Following standard transductive few-shot settings, our comprehensive experiments demonstrate that TIM outperforms state-of-the-art methods significantly across various datasets and networks, while used on top of a fixed feature extractor trained with simple cross-entropy on the base classes, without resorting to complex meta-learning schemes. It consistently brings between 2 % and 5 % improvement in accuracy over the best performing method, not only on all the well-established few-shot benchmarks but also on more challenging scenarios, with random tasks, domain shift and larger numbers of classes, as in the recently introduced META-DATASET. Our code is publicly available at https://github.com/mboudiaf/TIM. We also publicly release a standalone PyTorch implementation of META-DATASET, along with additional benchmarking results, at https://github.com/mboudiaf/pytorch-meta-dataset.

(2 min) Deep convolutional neural networks have been widely employed as an effective technique to handle complex and practical problems. However, one of the fundamental problems is the lack of formal methods to analyze their behavior. To address this challenge, we propose an approach to compute the exact reachable sets of a network given an input domain, where the reachable set is represented by the face lattice structure. Besides the computation of reachable sets, our approach is also capable of backtracking to the input domain given an output reachable set. Therefore, a full analysis of a network's behavior can be realized. In addition, an approach for fast analysis is also introduced, which conducts fast computation of reachable sets by considering selected sensitive neurons in each layer. The exact pixel-level reachability analysis method is evaluated on a CNN for the CIFAR10 dataset and compared to related works. The fast analysis method is evaluated over a CNN CIFAR10 dataset and VGG16 architecture for the ImageNet dataset.

(2 min) Behavioral scoring of research data is crucial for extracting domain-specific metrics but is bottlenecked on the ability to analyze enormous volumes of information using human labor. Deep learning is widely viewed as a key advancement to relieve this bottleneck. We identify one such domain, where deep learning can be leveraged to alleviate the process of manual scoring. Novelty preference paradigms have been widely used to study recognition memory in pigs, but analysis of these videos requires human intervention. We introduce a subset of such videos in the form of the 'Pig Novelty Preference Behavior' (PNPB) dataset that is fully annotated with pig actions and keypoints. In order to demonstrate the application of state-of-the-art action recognition models on this dataset, we compare LRCN, C3D, and TSM on the basis of various analytical metrics and discuss common pitfalls of the models. Our methods achieve an accuracy of 93% and a mean Average Precision of 96% in estimating piglet behavior. We open-source our code and annotated dataset at https://github.com/AIFARMS/NOR-behavior-recognition

(2 min) Image denoising is of great importance for medical imaging system, since it can improve image quality for disease diagnosis and downstream image analyses. In a variety of applications, dynamic imaging techniques are utilized to capture the time-varying features of the subject, where multiple images are acquired for the same subject at different time points. Although signal-to-noise ratio of each time frame is usually limited by the short acquisition time, the correlation among different time frames can be exploited to improve denoising results with shared information across time frames. With the success of neural networks in computer vision, supervised deep learning methods show prominent performance in single-image denoising, which rely on large datasets with clean-vs-noisy image pairs. Recently, several self-supervised deep denoising models have been proposed, achieving promising results without needing the pairwise ground truth of clean images. In the field of multi-image denoising, however, very few works have been done on extracting correlated information from multiple slices for denoising using self-supervised deep learning methods. In this work, we propose Deformed2Self, an end-to-end self-supervised deep learning framework for dynamic imaging denoising. It combines single-image and multi-image denoising to improve image quality and use a spatial transformer network to model motion between different slices. Further, it only requires a single noisy image with a few auxiliary observations at different time frames for training and inference. Evaluations on phantom and in vivo data with different noise statistics show that our method has comparable performance to other state-of-the-art unsupervised or self-supervised denoising methods and outperforms under high noise levels.

(2 min) One of the distinct features of this century has been the population of older adults which has been on a constant rise. Elderly people have several needs and requirements due to physical disabilities, cognitive issues, weakened memory and disorganized behavior, that they face with increasing age. The extent of these limitations also differs according to the varying diversities in elderly, which include age, gender, background, experience, skills, knowledge and so on. These varying needs and challenges with increasing age, limits abilities of older adults to perform Activities of Daily Living (ADLs) in an independent manner. To add to it, the shortage of caregivers creates a looming need for technology-based services for elderly people, to assist them in performing their daily routine tasks to sustain their independent living and active aging. To address these needs, this work consists of making three major contributions in this field. First, it provides a rather comprehensive review of assisted living technologies aimed at helping elderly people to perform ADLs. Second, the work discusses the challenges identified through this review, that currently exist in the context of implementation of assisted living services for elderly care in Smart Homes and Smart Cities. Finally, the work also outlines an approach for implementation, extension and integration of the existing works in this field for development of a much-needed framework that can provide personalized assistance and user-centered behavior interventions to elderly as per their varying and ever-changing needs.

(2 min) In deep learning area, large-scale image datasets bring a breakthrough in the success of object recognition and retrieval. Nowadays, as the embodiment of innovation, the diversity of the industrial goods is significantly larger, in which the incomplete multiview, multimodal and multilabel are different from the traditional dataset. In this paper, we introduce an industrial goods dataset, namely PatentNet, with numerous highly diverse, accurate and detailed annotations of industrial goods images, and corresponding texts. In PatentNet, the images and texts are sourced from design patent. Within over 6M images and corresponding texts of industrial goods labeled manually checked by professionals, PatentNet is the first ongoing industrial goods image database whose varieties are wider than industrial goods datasets used previously for benchmarking. PatentNet organizes millions of images into 32 classes and 219 subclasses based on the Locarno Classification Agreement. Through extensive experiments on image classification, image retrieval and incomplete multiview clustering, we demonstrate that our PatentNet is much more diverse, complex, and challenging, enjoying higher potentials than existing industrial image datasets. Furthermore, the characteristics of incomplete multiview, multimodal and multilabel in PatentNet are able to offer unparalleled opportunities in the artificial intelligence community and beyond.

(2 min) This paper jointly resolves two problems in vision transformer: i) the computation of Multi-Head Self-Attention (MHSA) has high computational/space complexity; ii) recent vision transformer networks are overly tuned for image classification, ignoring the difference between image classification (simple scenarios, more similar to NLP) and downstream scene understanding tasks (complicated scenarios, rich structural and contextual information). To this end, we note that pyramid pooling has been demonstrated to be effective in various vision tasks owing to its powerful context abstraction, and its natural property of spatial invariance is suitable to address the loss of structural information (problem ii)). Hence, we propose to adapt pyramid pooling to MHSA for alleviating its high requirement on computational resources (problem i)). In this way, this pooling-based MHSA can well address the above two problems and is thus flexible and powerful for downstream scene understanding tasks. Plugged with our pooling-based MHSA, we build a downstream-task-oriented transformer network, dubbed Pyramid Pooling Transformer (P2T). Extensive experiments demonstrate that, when applied P2T as the backbone network, it shows substantial superiority in various downstream scene understanding tasks such as semantic segmentation, object detection, instance segmentation, and visual saliency detection, compared to previous CNN- and transformer-based networks. The code will be released at https://github.com/yuhuan-wu/P2T. Note that this technical report will keep updating.

(2 min) Most modern deep learning-based multi-view 3D reconstruction techniques use RNNs or fusion modules to combine information from multiple images after encoding them. These two separate steps have loose connections and do not consider all available information while encoding each view. We propose LegoFormer, a transformer-based model that unifies object reconstruction under a single framework and parametrizes the reconstructed occupancy grid by its decomposition factors. This reformulation allows the prediction of an object as a set of independent structures then aggregated to obtain the final reconstruction. Experiments conducted on ShapeNet display the competitive performance of our network with respect to the state-of-the-art methods. We also demonstrate how the use of self-attention leads to increased interpretability of the model output.

(2 min) Over the years, accelerating neural networks with quantization has been widely studied. Unfortunately, prior efforts with diverse precisions (e.g., 1-bit weights and 2-bit activations) are usually restricted by limited precision support on GPUs (e.g., int1 and int4). To break such restrictions, we introduce the first Arbitrary Precision Neural Network framework (APNN-TC) to fully exploit quantization benefits on Ampere GPU Tensor Cores. Specifically, APNN-TC first incorporates a novel emulation algorithm to support arbitrary short bit-width computation with int1 compute primitives and XOR/AND Boolean operations. Second, APNN-TC integrates arbitrary precision layer designs to efficiently map our emulation algorithm to Tensor Cores with novel batching strategies and specialized memory organization. Third, APNN-TC embodies a novel arbitrary precision NN design to minimize memory access across layers and further improve performance. Extensive evaluations show that APNN-TC can achieve significant speedup over CUTLASS kernels and various NN models, such as ResNet and VGG.

(2 min) It is desirable to transfer the knowledge stored in a well-trained source model onto non-annotated target domain in the absence of source data. However, state-of-the-art methods for source free domain adaptation (SFDA) are subject to strict limits: 1) access to internal specifications of source models is a must; and 2) pseudo labels should be clean during self-training, making critical tasks relying on semantic segmentation unreliable. Aiming at these pitfalls, this study develops a domain adaptive solution to semantic segmentation with pseudo label rectification (namely \textit{PR-SFDA}), which operates in two phases: 1) \textit{Confidence-regularized unsupervised learning}: Maximum squares loss applies to regularize the target model to ensure the confidence in prediction; and 2) \textit{Noise-aware pseudo label learning}: Negative learning enables tolerance to noisy pseudo labels in training, meanwhile positive learning achieves fast convergence. Extensive experiments have been performed on domain adaptive semantic segmentation benchmark, \textit{GTA5 $\to$ Cityscapes}. Overall, \textit{PR-SFDA} achieves a performance of 49.0 mIoU, which is very close to that of the state-of-the-art counterparts. Note that the latter demand accesses to the source model's internal specifications, whereas the \textit{PR-SFDA} solution needs none as a sharp contrast.

(2 min) In this work, we propose a novel straightforward method for medical volume and sequence segmentation with limited annotations. To avert laborious annotating, the recent success of self-supervised learning(SSL) motivates the pre-training on unlabeled data. Despite its success, it is still challenging to adapt typical SSL methods to volume/sequence segmentation, due to their lack of mining on local semantic discrimination and rare exploitation on volume and sequence structures. Based on the continuity between slices/frames and the common spatial layout of organs across volumes/sequences, we introduced a novel bootstrap self-supervised representation learning method by leveraging the predictable possibility of neighboring slices. At the core of our method is a simple and straightforward dense self-supervision on the predictions of local representations and a strategy of predicting locals based on global context, which enables stable and reliable supervision for both global and local representation mining among volumes. Specifically, we first proposed an asymmetric network with an attention-guided predictor to enforce distance-specific prediction and supervision on slices within and across volumes/sequences. Secondly, we introduced a novel prototype-based foreground-background calibration module to enhance representation consistency. The two parts are trained jointly on labeled and unlabeled data. When evaluated on three benchmark datasets of medical volumes and sequences, our model outperforms existing methods with a large margin of 4.5\% DSC on ACDC, 1.7\% on Prostate, and 2.3\% on CAMUS. Intensive evaluations reveals the effectiveness and superiority of our method.

(2 min) This report describes the technical details of our submission to the EPIC-Kitchens 2021 Unsupervised Domain Adaptation Challenge for Action Recognition. The EPIC-Kitchens dataset is more difficult than other video domain adaptation datasets due to multi-tasks with more modalities. Firstly, to participate in the challenge, we employ a transformer to capture the spatial information from each modality. Secondly, we employ a temporal attention module to model temporal-wise inter-dependency. Thirdly, we employ the adversarial domain adaptation network to learn the general features between labeled source and unlabeled target domain. Finally, we incorporate multiple modalities to improve the performance by a three-stream network with late fusion. Our network achieves the comparable performance with the state-of-the-art baseline T$A^3$N and outperforms the baseline on top-1 accuracy for verb class and top-5 accuracies for all three tasks which are verb, noun and action. Under the team name xy9, our submission achieved 5th place in terms of top-1 accuracy for verb class and all top-5 accuracies.

(2 min) Deep neural networks are behind many of the recent successes in machine learning applications. However, these models can produce overconfident decisions while encountering out-of-distribution (OOD) examples or making a wrong prediction. This inconsistent predictive confidence limits the integration of independently-trained learning models into a larger system. This paper introduces separable concept learning framework to realistically measure the performance of classifiers in presence of OOD examples. In this setup, several instances of a classifier are trained on different parts of a partition of the set of classes. Later, the performance of the combination of these models is evaluated on a separate test set. Unlike current OOD detection techniques, this framework does not require auxiliary OOD datasets and does not separate classification from detection performance. Furthermore, we present a new strong baseline for more consistent predictive confidence in deep models, called fitted ensembles, where overconfident predictions are rectified by transformed versions of the original classification task. Fitted ensembles can naturally detect OOD examples without requiring auxiliary data by observing contradicting predictions among its components. Experiments on MNIST, SVHN, CIFAR-10/100, and ImageNet show fitted ensemble significantly outperform conventional ensembles on OOD examples and are possible to scale.

(2 min) Neural volume rendering became increasingly popular recently due to its success in synthesizing novel views of a scene from a sparse set of input images. So far, the geometry learned by neural volume rendering techniques was modeled using a generic density function. Furthermore, the geometry itself was extracted using an arbitrary level set of the density function leading to a noisy, often low fidelity reconstruction. The goal of this paper is to improve geometry representation and reconstruction in neural volume rendering. We achieve that by modeling the volume density as a function of the geometry. This is in contrast to previous work modeling the geometry as a function of the volume density. In more detail, we define the volume density function as Laplace's cumulative distribution function (CDF) applied to a signed distance function (SDF) representation. This simple density representation has three benefits: (i) it provides a useful inductive bias to the geometry learned in the neural volume rendering process; (ii) it facilitates a bound on the opacity approximation error, leading to an accurate sampling of the viewing ray. Accurate sampling is important to provide a precise coupling of geometry and radiance; and (iii) it allows efficient unsupervised disentanglement of shape and appearance in volume rendering. Applying this new density representation to challenging scene multiview datasets produced high quality geometry reconstructions, outperforming relevant baselines. Furthermore, switching shape and appearance between scenes is possible due to the disentanglement of the two.

(2 min) This article explores subspace clustering algorithms using CUR decompositions, and examines the effect of various hyperparameters in these algorithms on clustering performance on two real-world benchmark datasets, the Hopkins155 motion segmentation dataset and the Yale face dataset. Extensive experiments are done for a variety of sampling methods and oversampling parameters for these datasets, and some guidelines for parameter choices are given for practical applications.

(2 min) Many learning-based 3D shape semantic segmentation methods assign labels to shape atoms (e.g. points in a point cloud or faces in a mesh) with a single-pass approach trained in an end-to-end fashion. Such methods achieve impressive performance but require large amounts of labeled training data. This paradigm entangles two separable subproblems: (1) decomposing a shape into regions and (2) assigning semantic labels to these regions. We claim that disentangling these subproblems reduces the labeled data burden: (1) region decomposition requires no semantic labels and could be performed in an unsupervised fashion, and (2) labeling shape regions instead of atoms results in a smaller search space and should be learnable with less labeled training data. In this paper, we investigate this second claim by presenting the Neurally-Guided Shape Parser (NGSP), a method that learns how to assign semantic labels to regions of an over-segmented 3D shape. We solve this problem via MAP inference, modeling the posterior probability of a labeling assignment conditioned on an input shape. We employ a Monte Carlo importance sampling approach guided by a neural proposal network, a search-based approach made feasible by assuming the input shape is decomposed into discrete regions. We evaluate NGSP on the task of hierarchical semantic segmentation on manufactured 3D shapes from PartNet. We find that NGSP delivers significant performance improvements over baselines that learn to label shape atoms and then aggregate predictions for each shape region, especially in low-data regimes. Finally, we demonstrate that NGSP is robust to region granularity, as it maintains strong segmentation performance even as the regions undergo significant corruption.

(2 min) Detailed analysis of seizure semiology, the symptoms and signs which occur during a seizure, is critical for management of epilepsy patients. Inter-rater reliability using qualitative visual analysis is often poor for semiological features. Therefore, automatic and quantitative analysis of video-recorded seizures is needed for objective assessment. We present GESTURES, a novel architecture combining convolutional neural networks (CNNs) and recurrent neural networks (RNNs) to learn deep representations of arbitrarily long videos of epileptic seizures. We use a spatiotemporal CNN (STCNN) pre-trained on large human action recognition (HAR) datasets to extract features from short snippets (approx. 0.5 s) sampled from seizure videos. We then train an RNN to learn seizure-level representations from the sequence of features. We curated a dataset of seizure videos from 68 patients and evaluated GESTURES on its ability to classify seizures into focal onset seizures (FOSs) (N = 106) vs. focal to bilateral tonic-clonic seizures (TCSs) (N = 77), obtaining an accuracy of 98.9% using bidirectional long short-term memory (BLSTM) units. We demonstrate that an STCNN trained on a HAR dataset can be used in combination with an RNN to accurately represent arbitrarily long videos of seizures. GESTURES can provide accurate seizure classification by modeling sequences of semiologies.

(2 min) Transmission electron microscopy (TEM) is one of the primary tools to show microstructural characterization of materials as well as film thickness. However, manual determination of film thickness from TEM images is time-consuming as well as subjective, especially when the films in question are very thin and the need for measurement precision is very high. Such is the case for head overcoat (HOC) thickness measurements in the magnetic hard disk drive industry. It is therefore necessary to develop software to automatically measure HOC thickness. In this paper, for the first time, we propose a HOC layer segmentation method using NASNet-Large as an encoder and then followed by a decoder architecture, which is one of the most commonly used architectures in deep learning for image segmentation. To further improve segmentation results, we are the first to propose a post-processing layer to remove irrelevant portions in the segmentation result. To measure the thickness of the segmented HOC layer, we propose a regressive convolutional neural network (RCNN) model as well as orthogonal thickness calculation methods. Experimental results demonstrate a higher dice score for our model which has lower mean squared error and outperforms current state-of-the-art manual measurement.

(2 min) Deep learning has recently been applied to optical music recognition (OMR). However, currently OMR processing from various sheet music images still lacks precision to be widely applicable. Here, we present an MMdA (Measure-based Multimodal deep learning (DL)-driven Assembly) method allowing for end-to-end OMR processing from various images including inclined photo images. Using this method, measures are extracted by a deep learning model, aligned, and resized to be used for inference of given musical symbol components by using multiple deep learning models in sequence or in parallel. Use of each standardized measure enables efficient training of the models and accurate adjustment of five staff lines in each measure. Multiple musical symbol component category models with a small number of feature types can represent a diverse set of notes and other musical symbols including chords. This MMdA method provides a solution to end-to-end OMR processing with precision.

(2 min) We focus on exploring various approaches of Zero-Shot Learning (ZSL) and their explainability for a challenging yet important supervised learning task notorious for training data scarcity, i.e. Depression Symptoms Detection (DSD) from text. We start with a comprehensive synthesis of different components of our ZSL modeling and analysis of our ground truth samples and Depression symptom clues curation process with the help of a practicing clinician. We next analyze the accuracy of various state-of-the-art ZSL models and their potential enhancements for our task. Further, we sketch a framework for the use of ZSL for hierarchical text-based explanation mechanism, which we call, Syntax Tree-Guided Semantic Explanation (STEP). Finally, we summarize experiments from which we conclude that we can use ZSL models and achieve reasonable accuracy and explainability, measured by a proposed Explainability Index (EI). This work is, to our knowledge, the first work to exhaustively explore the efficacy of ZSL models for DSD task, both in terms of accuracy and explainability.

(2 min) We present a graph-based approach for the data management tasks and the efficient operation of a system for session-based next-item recommendations. The proposed method can collect data continuously and incrementally from an ecommerce web site, thus seemingly prepare the necessary data infrastructure for the recommendation algorithm to operate without any excessive training phase. Our work aims at developing a recommender method that represents a balance between data processing and management efficiency requirements and the effectiveness of the recommendations produced. We use the Neo4j graph database to implement a prototype of such a system. Furthermore, we use an industry dataset corresponding to a typical e-commerce session-based scenario, and we report on experiments using our graph-based approach and other state-of-the-art machine learning and deep learning methods.

(2 min) Machine learning models for the ad-hoc retrieval of documents and passages have recently shown impressive improvements due to better language understanding using large pre-trained language models. However, these over-parameterized models are inherently non-interpretable and do not provide any information on the parts of the documents that were used to arrive at a certain prediction. In this paper we introduce the select and rank paradigm for document ranking, where interpretability is explicitly ensured when scoring longer documents. Specifically, we first select sentences in a document based on the input query and then predict the query-document score based only on the selected sentences, acting as an explanation. We treat sentence selection as a latent variable trained jointly with the ranker from the final output. We conduct extensive experiments to demonstrate that our inherently interpretable select-and-rank approach is competitive in comparison to other state-of-the-art methods and sometimes even outperforms them. This is due to our novel end-to-end training approach based on weighted reservoir sampling that manages to train the selector despite the stochastic sentence selection. We also show that our sentence selection approach can be used to provide explanations for models that operate on only parts of the document, such as BERT.

(2 min) Word embeddings (e.g., word2vec) have been applied successfully to eCommerce products through~\textit{prod2vec}. Inspired by the recent performance improvements on several NLP tasks brought by contextualized embeddings, we propose to transfer BERT-like architectures to eCommerce: our model -- ~\textit{Prod2BERT} -- is trained to generate representations of products through masked session modeling. Through extensive experiments over multiple shops, different tasks, and a range of design choices, we systematically compare the accuracy of~\textit{Prod2BERT} and~\textit{prod2vec} embeddings: while~\textit{Prod2BERT} is found to be superior in several scenarios, we highlight the importance of resources and hyperparameters in the best performing models. Finally, we provide guidelines to practitioners for training embeddings under a variety of computational and data constraints.

(2 min) The insights revealed from process mining heavily rely on the quality of event logs. Activities extracted from healthcare information systems with the free-text nature may lead to inconsistent labels. Such inconsistency would then lead to redundancy of activity labels, which refer to labels that have different syntax but share the same behaviours. The identifications of these labels from data-driven process discovery are difficult and rely heavily on resource-intensive human review. Existing work achieves low accuracy either redundant activity labels are in low occurrence frequency or the existence of numerical data values as attributes in event logs. However, these phenomena are commonly observed in healthcare information systems. In this paper, we propose an approach to detect redundant activity labels using control-flow relations and numerical data values from event logs. Natural Language Processing is also integrated into our method to assess semantic similarity between labels, which provides users with additional insights. We have evaluated our approach through synthetic logs generated from the real-life Sepsis log and a case study using the MIMIC-III data set. The results demonstrate that our approach can successfully detect redundant activity labels. This approach can add value to the preprocessing step to generate more representative event logs for process mining tasks in the healthcare domain.

(2 min) In today's academic publishing model, especially in Computer Science, conferences commonly constitute the main platforms for releasing the latest peer-reviewed advancements in their respective fields. However, choosing a suitable academic venue for publishing one's research can represent a challenging task considering the plethora of available conferences, particularly for those at the start of their academic careers, or for those seeking to publish outside of their usual domain. In this paper, we propose GraphConfRec, a conference recommender system which combines SciGraph and graph neural networks, to infer suggestions based not only on title and abstract, but also on co-authorship and citation relationships. GraphConfRec achieves a recall@10 of up to 0.580 and a MAP of up to 0.336 with a graph attention network-based recommendation model. A user study with 25 subjects supports the positive results.

(2 min) This paper proposes new acoustic feature signatures based on the multiscale fractal dimension (MFD), which are robust against the diversity of environmental sounds, for the content-based similarity search. The diversity of sound sources and acoustic compositions is a typical feature of environmental sounds. Several acoustic features have been proposed for environmental sounds. Among them is the widely-used Mel-Frequency Cepstral Coefficients (MFCCs), which describes frequency-domain features. However, in addition to these features in the frequency domain, environmental sounds have other important features in the time domain with various time scales. In our previous paper, we proposed enhanced multiscale fractal dimension signature (EMFD) for environmental sounds. This paper extends EMFD by using the kernel density estimation method, which results in better performance of the similarity search tasks. Furthermore, it newly proposes another acoustic feature signature based on MFD, namely very-long-range multiscale fractal dimension signature (MFD-VL). The MFD-VL signature describes several features of the time-varying envelope for long periods of time. The MFD-VL signature has stability and robustness against background noise and small fluctuations in the parameters of sound sources, which are produced in field recordings. We discuss the effectiveness of these signatures in the similarity sound search by comparing with acoustic features proposed in the DCASE 2018 challenges. Due to the unique descriptiveness of our proposed signatures, we confirmed the signatures are effective when they are used with other acoustic features.

(2 min) Automatic processing of bibliographic data becomes very important in digital libraries, data science and machine learning due to its importance in keeping pace with the significant increase of published papers every year from one side and to the inherent challenges from the other side. This processing has several aspects including but not limited to I) Automatic extraction of references from PDF documents, II) Building an accurate citation graph, III) Author name disambiguation, etc. Bibliographic data is heterogeneous by nature and occurs in both structured (e.g. citation graph) and unstructured (e.g. publications) formats. Therefore, it requires data science and machine learning techniques to be processed and analysed. Here we introduce BiblioDAP'21: The 1st Workshop on Bibliographic Data Analysis and Processing.

(2 min) One of the key challenges in Sequential Recommendation (SR) is how to extract and represent user preferences. Traditional SR methods rely on the next item as the supervision signal to guide preference extraction and representation. We propose a novel learning strategy, named preference editing. The idea is to force the SR model to discriminate the common and unique preferences in different sequences of interactions between users and the recommender system. By doing so, the SR model is able to learn how to identify common and unique user preferences, and thereby do better user preference extraction and representation. We propose a transformer based SR model, named MrTransformer (Multi-preference Transformer), that concatenates some special tokens in front of the sequence to represent multiple user preferences and makes sure they capture different aspects through a preference coverage mechanism. Then, we devise a preference editing-based self-supervised learning mechanism for training MrTransformer which contains two main operations: preference separation and preference recombination. The former separates the common and unique user preferences for a given pair of sequences. The latter swaps the common preferences to obtain recombined user preferences for each sequence. Based on the preference separation and preference recombination operations, we define two types of SSL loss that require that the recombined preferences are similar to the original ones, and the common preferences are close to each other. We carry out extensive experiments on two benchmark datasets. MrTransformer with preference editing significantly outperforms state-of-the-art SR methods in terms of Recall, MRR and NDCG. We find that long sequences whose user preferences are harder to extract and represent benefit most from preference editing.

(2 min) Multi-armed bandits (MAB) provide a principled online learning approach to attain the balance between exploration and exploitation. Due to the superior performance and low feedback learning without the learning to act in multiple situations, Multi-armed Bandits drawing widespread attention in applications ranging such as recommender systems. Likewise, within the recommender system, collaborative filtering (CF) is arguably the earliest and most influential method in the recommender system. Crucially, new users and an ever-changing pool of recommended items are the challenges that recommender systems need to address. For collaborative filtering, the classical method is training the model offline, then perform the online testing, but this approach can no longer handle the dynamic changes in user preferences which is the so-called cold start. So how to effectively recommend items to users in the absence of effective information? To address the aforementioned problems, a multi-armed bandit based collaborative filtering recommender system has been proposed, named BanditMF. BanditMF is designed to address two challenges in the multi-armed bandits algorithm and collaborative filtering: (1) how to solve the cold start problem for collaborative filtering under the condition of scarcity of valid information, (2) how to solve the sub-optimal problem of bandit algorithms in strong social relations domains caused by independently estimating unknown parameters associated with each user and ignoring correlations between users.

(2 min) We focus on exploring various approaches of Zero-Shot Learning (ZSL) and their explainability for a challenging yet important supervised learning task notorious for training data scarcity, i.e. Depression Symptoms Detection (DSD) from text. We start with a comprehensive synthesis of different components of our ZSL modeling and analysis of our ground truth samples and Depression symptom clues curation process with the help of a practicing clinician. We next analyze the accuracy of various state-of-the-art ZSL models and their potential enhancements for our task. Further, we sketch a framework for the use of ZSL for hierarchical text-based explanation mechanism, which we call, Syntax Tree-Guided Semantic Explanation (STEP). Finally, we summarize experiments from which we conclude that we can use ZSL models and achieve reasonable accuracy and explainability, measured by a proposed Explainability Index (EI). This work is, to our knowledge, the first work to exhaustively explore the efficacy of ZSL models for DSD task, both in terms of accuracy and explainability.

(2 min) We present a fast and feature-complete differentiable physics engine, Nimble (nimblephysics.org), that supports Lagrangian dynamics and hard contact constraints for articulated rigid body simulation. Our differentiable physics engine offers a complete set of features that are typically only available in non-differentiable physics simulators commonly used by robotics applications. We solve contact constraints precisely using linear complementarity problems (LCPs). We present efficient and novel analytical gradients through the LCP formulation of inelastic contact that exploit the sparsity of the LCP solution. We support complex contact geometry, and gradients approximating continuous-time elastic collision. We also introduce a novel method to compute complementarity-aware gradients that help downstream optimization tasks avoid stalling in saddle points. We show that an implementation of this combination in an existing physics engine (DART) is capable of a 87x single-core speedup over finite-differencing in computing analytical Jacobians for a single timestep, while preserving all the expressiveness of original DART.

(2 min) We propose, implement, and evaluate a new algorithm for releasing answers to very large numbers of statistical queries like $k$-way marginals, subject to differential privacy. Our algorithm makes adaptive use of a continuous relaxation of the Projection Mechanism, which answers queries on the private dataset using simple perturbation, and then attempts to find the synthetic dataset that most closely matches the noisy answers. We use a continuous relaxation of the synthetic dataset domain which makes the projection loss differentiable, and allows us to use efficient ML optimization techniques and tooling. Rather than answering all queries up front, we make judicious use of our privacy budget by iteratively and adaptively finding queries for which our (relaxed) synthetic data has high error, and then repeating the projection. We perform extensive experimental evaluations across a range of parameters and datasets, and find that our method outperforms existing algorithms in many cases, especially when the privacy budget is small or the query class is large.

(2 min) While common image object detection tasks focus on bounding boxes or segmentation masks as object representations, we consider the problem of finding objects based on four arbitrary vertices. We propose a novel model, named TetraPackNet, to tackle this problem. TetraPackNet is based on CornerNet and uses similar algorithms and ideas. It is designated for applications requiring high-accuracy detection of regularly shaped objects, which is the case in the logistics use-case of packaging structure recognition. We evaluate our model on our specific real-world dataset for this use-case. Baselined against a previous solution, consisting of a Mask R-CNN model and suitable post-processing steps, TetraPackNet achieves superior results (9% higher in accuracy) in the sub-task of four-corner based transport unit side detection.

(2 min) Fine-scale simulation of complex systems governed by multiscale partial differential equations (PDEs) is computationally expensive and various multiscale methods have been developed for addressing such problems. In addition, it is challenging to develop accurate surrogate and uncertainty quantification models for high-dimensional problems governed by stochastic multiscale PDEs using limited training data. In this work to address these challenges, we introduce a novel hybrid deep-learning and multiscale approach for stochastic multiscale PDEs with limited training data. For demonstration purposes, we focus on a porous media flow problem. We use an image-to-image supervised deep learning model to learn the mapping between the input permeability field and the multiscale basis functions. We introduce a Bayesian approach to this hybrid framework to allow us to perform uncertainty quantification and propagation tasks. The performance of this hybrid approach is evaluated with varying intrinsic dimensionality of the permeability field. Numerical results indicate that the hybrid network can efficiently predict well for high-dimensional inputs.

(2 min) We investigate the complexity and performance of recurrent neural network (RNN) models as post-processing units for the compensation of fibre nonlinearities in digital coherent systems carrying polarization multiplexed 16-QAM and 32-QAM signals. We evaluate three bi-directional RNN models, namely the bi-LSTM, bi-GRU and bi-Vanilla-RNN and show that all of them are promising nonlinearity compensators especially in dispersion unmanaged systems. Our simulations show that during inference the three models provide similar compensation performance, therefore in real-life systems the simplest scheme based on Vanilla-RNN units should be preferred. We compare bi-Vanilla-RNN with Volterra nonlinear equalizers and exhibit its superiority both in terms of performance and complexity, thus highlighting that RNN processing is a very promising pathway for the upgrade of long-haul optical communication systems utilizing coherent detection.

(2 min) There has been a recent resurgence of interest in explainable artificial intelligence (XAI) that aims to reduce the opaqueness of AI-based decision-making systems, allowing humans to scrutinize and trust them. Prior work in this context has focused on the attribution of responsibility for an algorithm's decisions to its inputs wherein responsibility is typically approached as a purely associational concept. In this paper, we propose a principled causality-based approach for explaining black-box decision-making systems that addresses limitations of existing methods in XAI. At the core of our framework lies probabilistic contrastive counterfactuals, a concept that can be traced back to philosophical, cognitive, and social foundations of theories on how humans generate and select explanations. We show how such counterfactuals can quantify the direct and indirect influences of a variable on decisions made by an algorithm, and provide actionable recourse for individuals negatively affected by the algorithm's decision. Unlike prior work, our system, LEWIS: (1)can compute provably effective explanations and recourse at local, global and contextual levels (2)is designed to work with users with varying levels of background knowledge of the underlying causal model and (3)makes no assumptions about the internals of an algorithmic system except for the availability of its input-output data. We empirically evaluate LEWIS on three real-world datasets and show that it generates human-understandable explanations that improve upon state-of-the-art approaches in XAI, including the popular LIME and SHAP. Experiments on synthetic data further demonstrate the correctness of LEWIS's explanations and the scalability of its recourse algorithm.

(2 min) End-to-end models with auto-regressive decoders have shown impressive results for automatic speech recognition (ASR). These models formulate the sequence-level probability as a product of the conditional probabilities of all individual tokens given their histories. However, the performance of locally normalised models can be sub-optimal because of factors such as exposure bias. Consequently, the model distribution differs from the underlying data distribution. In this paper, the residual energy-based model (R-EBM) is proposed to complement the auto-regressive ASR model to close the gap between the two distributions. Meanwhile, R-EBMs can also be regarded as utterance-level confidence estimators, which may benefit many downstream tasks. Experiments on a 100hr LibriSpeech dataset show that R-EBMs can reduce the word error rates (WERs) by 8.2%/6.7% while improving areas under precision-recall curves of confidence scores by 12.6%/28.4% on test-clean/test-other sets. Furthermore, on a state-of-the-art model using self-supervised learning (wav2vec 2.0), R-EBMs still significantly improves both the WER and confidence estimation performance.

(2 min) Assessing the complexity of functions computed by a neural network helps us understand how the network will learn and generalize. One natural measure of complexity is how the network distorts length - if the network takes a unit-length curve as input, what is the length of the resulting curve of outputs? It has been widely believed that this length grows exponentially in network depth. We prove that in fact this is not the case: the expected length distortion does not grow with depth, and indeed shrinks slightly, for ReLU networks with standard random initialization. We also generalize this result by proving upper bounds both for higher moments of the length distortion and for the distortion of higher-dimensional volumes. These theoretical results are corroborated by our experiments.

(2 min) We consider measurement-feedback control in linear dynamical systems from the perspective of regret minimization. Unlike most prior work in this area, we focus on the problem of designing an online controller which competes with the optimal dynamic sequence of control actions selected in hindsight, instead of the best controller in some specific class of controllers. This formulation of regret is attractive when the environment changes over time and no single controller achieves good performance over the entire time horizon. We show that in the measurement-feedback setting, unlike in the full-information setting, there is no single offline controller which outperforms every other offline controller on every disturbance, and propose a new $H_2$-optimal offline controller as a benchmark for the online controller to compete against. We show that the corresponding regret-optimal online controller can be found via a novel reduction to the classical Nehari problem from robust control and present a tight data-dependent bound on its regret.

(2 min) Multi-armed bandits (MAB) provide a principled online learning approach to attain the balance between exploration and exploitation. Due to the superior performance and low feedback learning without the learning to act in multiple situations, Multi-armed Bandits drawing widespread attention in applications ranging such as recommender systems. Likewise, within the recommender system, collaborative filtering (CF) is arguably the earliest and most influential method in the recommender system. Crucially, new users and an ever-changing pool of recommended items are the challenges that recommender systems need to address. For collaborative filtering, the classical method is training the model offline, then perform the online testing, but this approach can no longer handle the dynamic changes in user preferences which is the so-called cold start. So how to effectively recommend items to users in the absence of effective information? To address the aforementioned problems, a multi-armed bandit based collaborative filtering recommender system has been proposed, named BanditMF. BanditMF is designed to address two challenges in the multi-armed bandits algorithm and collaborative filtering: (1) how to solve the cold start problem for collaborative filtering under the condition of scarcity of valid information, (2) how to solve the sub-optimal problem of bandit algorithms in strong social relations domains caused by independently estimating unknown parameters associated with each user and ignoring correlations between users.

(2 min) A popular way to create detailed yet easily controllable 3D shapes is via procedural modeling, i.e. generating geometry using programs. Such programs consist of a series of instructions along with their associated parameter values. To fully realize the benefits of this representation, a shape program should be compact and only expose degrees of freedom that allow for meaningful manipulation of output geometry. One way to achieve this goal is to design higher-level macro operators that, when executed, expand into a series of commands from the base shape modeling language. However, manually authoring such macros, much like shape programs themselves, is difficult and largely restricted to domain experts. In this paper, we present ShapeMOD, an algorithm for automatically discovering macros that are useful across large datasets of 3D shape programs. ShapeMOD operates on shape programs expressed in an imperative, statement-based language. It is designed to discover macros that make programs more compact by minimizing the number of function calls and free parameters required to represent an input shape collection. We run ShapeMOD on multiple collections of programs expressed in a domain-specific language for 3D shape structures. We show that it automatically discovers a concise set of macros that abstract out common structural and parametric patterns that generalize over large shape collections. We also demonstrate that the macros found by ShapeMOD improve performance on downstream tasks including shape generative modeling and inferring programs from point clouds. Finally, we conduct a user study that indicates that ShapeMOD's discovered macros make interactive shape editing more efficient.

(2 min) Neural Architecture Search (NAS) often trains and evaluates a large number of architectures. Recent predictor-based NAS approaches attempt to address such heavy computation costs with two key steps: sampling some architecture-performance pairs and fitting a proxy accuracy predictor. Given limited samples, these predictors, however, are far from accurate to locate top architectures due to the difficulty of fitting the huge search space. This paper reflects on a simple yet crucial question: if our final goal is to find the best architecture, do we really need to model the whole space well?. We propose a paradigm shift from fitting the whole architecture space using one strong predictor, to progressively fitting a search path towards the high-performance sub-space through a set of weaker predictors. As a key property of the proposed weak predictors, their probabilities of sampling better architectures keep increasing. Hence we only sample a few well-performed architectures guided by the previously learned predictor and estimate a new better weak predictor. This embarrassingly easy framework produces coarse-to-fine iteration to refine the ranking of sampling space gradually. Extensive experiments demonstrate that our method costs fewer samples to find top-performance architectures on NAS-Bench-101 and NAS-Bench-201, as well as achieves the state-of-the-art ImageNet performance on the NASNet search space. In particular, compared to state-of-the-art (SOTA) predictor-based NAS methods, WeakNAS outperforms all of them with notable margins, e.g., requiring at least 7.5x less samples to find global optimal on NAS-Bench-101; and WeakNAS can also absorb them for further performance boost. We further strike the new SOTA result of 81.3% in the ImageNet MobileNet Search Space. The code is available at https://github.com/VITA-Group/WeakNAS.

(2 min) This paper deals with an important subject in classification problems addressed by machine learning techniques: the evaluation of the influence of each of the features on the classification of individuals. Specifically, a measure of that influence is introduced using the Shapley value of cooperative games. In addition, an axiomatic characterisation of the proposed measure is provided based on properties of efficiency and balanced contributions. Furthermore, some experiments have been designed in order to validate the appropriate performance of such measure. Finally, the methodology introduced is applied to a sample of COVID-19 patients to study the influence of certain demographic or risk factors on various events of interest related to the evolution of the disease.

(2 min) In many applications, it is desirable that a classifier not only makes accurate predictions, but also outputs calibrated posterior probabilities. However, many existing classifiers, especially deep neural network classifiers, tend to be uncalibrated. Post-hoc calibration is a technique to recalibrate a model by learning a calibration map. Existing approaches mostly focus on constructing calibration maps with low calibration errors, however, this quality is inadequate for a calibrator being useful. In this paper, we introduce two constraints that are worth consideration in designing a calibration map for post-hoc calibration. Then we present Meta-Cal, which is built from a base calibrator and a ranking model. Under some mild assumptions, two high-probability bounds are given with respect to these constraints. Empirical results on CIFAR-10, CIFAR-100 and ImageNet and a range of popular network architectures show our proposed method significantly outperforms the current state of the art for post-hoc multi-class classification calibration.

(2 min) We investigate the benefits of feature selection, nonlinear modelling and online learning when forecasting in financial time series. We consider the sequential and continual learning sub-genres of online learning. The experiments we conduct show that there is a benefit to online transfer learning, in the form of radial basis function networks, beyond the sequential updating of recursive least-squares models. We show that the radial basis function networks, which make use of clustering algorithms to construct a kernel Gram matrix, are more beneficial than treating each training vector as separate basis functions, as occurs with kernel Ridge regression. We demonstrate quantitative procedures to determine the very structure of the radial basis function networks. Finally, we conduct experiments on the log returns of financial time series and show that the online learning models, particularly the radial basis function networks, are able to outperform a random walk baseline, whereas the offline learning models struggle to do so.

(2 min) Despite the empirical success of neural architecture search (NAS) in deep learning applications, the optimality, reproducibility and cost of NAS schemes remain hard to assess. In this paper, we propose Generative Adversarial NAS (GA-NAS) with theoretically provable convergence guarantees, promoting stability and reproducibility in neural architecture search. Inspired by importance sampling, GA-NAS iteratively fits a generator to previously discovered top architectures, thus increasingly focusing on important parts of a large search space. Furthermore, we propose an efficient adversarial learning approach, where the generator is trained by reinforcement learning based on rewards provided by a discriminator, thus being able to explore the search space without evaluating a large number of architectures. Extensive experiments show that GA-NAS beats the best published results under several cases on three public NAS benchmarks. In the meantime, GA-NAS can handle ad-hoc search constraints and search spaces. We show that GA-NAS can be used to improve already optimized baselines found by other NAS methods, including EfficientNet and ProxylessNAS, in terms of ImageNet accuracy or the number of parameters, in their original search space.

(2 min) Splitting network computations between the edge device and a server enables low edge-compute inference of neural networks but might expose sensitive information about the test query to the server. To address this problem, existing techniques train the model to minimize information leakage for a given set of sensitive attributes. In practice, however, the test queries might contain attributes that are not foreseen during training. We propose instead an unsupervised obfuscation method to discard the information irrelevant to the main task. We formulate the problem via an information theoretical framework and derive an analytical solution for a given distortion to the model output. In our method, the edge device runs the model up to a split layer determined based on its computational capacity. It then obfuscates the obtained feature vector based on the first layer of the server model by removing the components in the null space as well as the low-energy components of the remaining signal. Our experimental results show that our method outperforms existing techniques in removing the information of the irrelevant attributes and maintaining the accuracy on the target label. We also show that our method reduces the communication cost and incurs only a small computational overhead.

(2 min) This paper proposes a machine learning method based on the Extra Trees (ET) algorithm for forecasting Significant Wave Heights in oceanic waters. To derive multiple features from the CDIP buoys, which make point measurements, we first nowcast various parameters and then forecast them at 30-min intervals. The proposed algorithm has Scatter Index (SI), Bias, Correlation Coefficient, Root Mean Squared Error (RMSE) of 0.130, -0.002, 0.97, and 0.14, respectively, for one day ahead prediction and 0.110, -0.001, 0.98, and 0.122, respectively, for 14-day ahead prediction on the testing dataset. While other state-of-the-art methods can only forecast up to 120 hours ahead, we extend it further to 14 days. Our proposed setup includes spectral features, hv-block cross-validation, and stringent QC criteria. The proposed algorithm performs significantly better than the state-of-the-art methods commonly used for significant wave height forecasting for one-day ahead prediction. Moreover, the improved performance of the proposed machine learning method compared to the numerical methods shows that this performance can be extended to even longer periods allowing for early prediction of significant wave heights in oceanic waters.

(2 min) Recently, visual Transformer (ViT) and its following works abandon the convolution and exploit the self-attention operation, attaining a comparable or even higher accuracy than CNNs. More recently, MLP-Mixer abandons both the convolution and the self-attention operation, proposing an architecture containing only MLP layers. To achieve cross-patch communications, it devises an additional token-mixing MLP besides the channel-mixing MLP. It achieves promising results when training on an extremely large-scale dataset. But it cannot achieve as outstanding performance as its CNN and ViT counterparts when training on medium-scale datasets such as ImageNet1K and ImageNet21K. The performance drop of MLP-Mixer motivates us to rethink the token-mixing MLP. We discover that the token-mixing MLP is a variant of the depthwise convolution with a global reception field and spatial-specific configuration. But the global reception field and the spatial-specific property make token-mixing MLP prone to over-fitting. In this paper, we propose a novel pure MLP architecture, spatial-shift MLP (S$^2$-MLP). Different from MLP-Mixer, our S$^2$-MLP only contains channel-mixing MLP. We utilize a spatial-shift operation for communications between patches. It has a local reception field and is spatial-agnostic. It is parameter-free and efficient for computation. The proposed S$^2$-MLP attains higher recognition accuracy than MLP-Mixer when training on ImageNet-1K dataset. Meanwhile, S$^2$-MLP accomplishes as excellent performance as ViT on ImageNet-1K dataset with considerably simpler architecture and fewer FLOPs and parameters.

(2 min) Being able to efficiently and accurately select the top-$k$ elements with differential privacy is an integral component of various private data analysis tasks. In this paper, we present the oneshot Laplace mechanism, which generalizes the well-known Report Noisy Max mechanism to reporting noisy top-$k$ elements. We show that the oneshot Laplace mechanism with a noise level of $\widetilde{O}(\sqrt{k}/\eps)$ is approximately differentially private. Compared to the previous peeling approach of running Report Noisy Max $k$ times, the oneshot Laplace mechanism only adds noises and computes the top $k$ elements once, hence much more efficient for large $k$. In addition, our proof of privacy relies on a novel coupling technique that bypasses the use of composition theorems. Finally, we present a novel application of efficient top-$k$ selection in the classical problem of ranking from pairwise comparisons.

(2 min) Deep generative models have recently demonstrated the ability to synthesize photorealistic images of human faces with novel identities. A key challenge to the wide applicability of such techniques is to provide independent control over semantically meaningful parameters: appearance, head pose, face shape, and facial expressions. In this paper, we propose VariTex - to the best of our knowledge the first method that learns a variational latent feature space of neural face textures, which allows sampling of novel identities. We combine this generative model with a parametric face model and gain explicit control over head pose and facial expressions. To generate images of complete human heads, we propose an additive decoder that generates plausible additional details such as hair. A novel training scheme enforces a pose independent latent space and in consequence, allows learning of a one-to-many mapping between latent codes and pose-conditioned exterior regions. The resulting method can generate geometrically consistent images of novel identities allowing fine-grained control over head pose, face shape, and facial expressions, facilitating a broad range of downstream tasks, like sampling novel identities, re-posing, expression transfer, and more.

(2 min) Learning a sequence of tasks without access to i.i.d. observations is a widely studied form of continual learning (CL) that remains challenging. In principle, Bayesian learning directly applies to this setting, since recursive and one-off Bayesian updates yield the same result. In practice, however, recursive updating often leads to poor trade-off solutions across tasks because approximate inference is necessary for most models of interest. Here, we describe an alternative Bayesian approach where task-conditioned parameter distributions are continually inferred from data. We offer a practical deep learning implementation of our framework based on probabilistic task-conditioned hypernetworks, an approach we term "posterior meta-replay". Experiments on standard benchmarks show that our probabilistic hypernetworks compress sequences of posterior parameter distributions with virtually no forgetting. We obtain considerable performance gains compared to existing Bayesian CL methods, and identify task inference as our major limiting factor. This limitation has several causes that are independent of the considered sequential setting, opening up new avenues for progress in CL.

(2 min) A deep learning approach is proposed to detect data and system anomalies using high-resolution continuous point-on-wave (CPOW) or phasor measurements. Both the anomaly and anomaly-free measurement models are assumed to have unknown temporal dependencies and probability distributions. Historical training samples are assumed for the anomaly-free model, while no training samples are available for the anomaly measurements. By transforming the anomaly-free observations into uniform independent and identically distributed sequences via a generative adversarial network, the proposed approach deploys a uniformity test for anomaly detection at the sensor level. A distributed detection scheme that combines sensor level detections at the control center is also proposed that combines local detections to form more reliable detections. Numerical results demonstrate significant improvement over the state-of-the-art solutions for various bad-data cases using real and synthetic CPOW and PMU data sets.

(2 min) OpenML is an online platform for open science collaboration in machine learning, used to share datasets and results of machine learning experiments. In this paper we introduce OpenML-Python, a client API for Python, opening up the OpenML platform for a wide range of Python-based tools. It provides easy access to all datasets, tasks and experiments on OpenML from within Python. It also provides functionality to conduct machine learning experiments, upload the results to OpenML, and reproduce results which are stored on OpenML. Furthermore, it comes with a scikit-learn plugin and a plugin mechanism to easily integrate other machine learning libraries written in Python into the OpenML ecosystem. Source code and documentation is available at https://github.com/openml/openml-python/.

(2 min) The outbreak of COVID-19 Disease due to the novel coronavirus has caused a shortage of medical resources. To aid and accelerate the diagnosis process, automatic diagnosis of COVID-19 via deep learning models has recently been explored by researchers across the world. While different data-driven deep learning models have been developed to mitigate the diagnosis of COVID-19, the data itself is still scarce due to patient privacy concerns. Federated Learning (FL) is a natural solution because it allows different organizations to cooperatively learn an effective deep learning model without sharing raw data. However, recent studies show that FL still lacks privacy protection and may cause data leakage. We investigate this challenging problem by proposing a simple yet effective algorithm, named \textbf{F}ederated \textbf{L}earning \textbf{o}n Medical Datasets using \textbf{P}artial Networks (FLOP), that shares only a partial model between the server and clients. Extensive experiments on benchmark data and real-world healthcare tasks show that our approach achieves comparable or better performance while reducing the privacy and security risks. Of particular interest, we conduct experiments on the COVID-19 dataset and find that our FLOP algorithm can allow different hospitals to collaboratively and effectively train a partially shared model without sharing local patients' data.

(2 min) Attribution methods have been shown as promising approaches for identifying key features that led to learned model predictions. While most existing attribution methods rely on a baseline input for performing feature perturbations, limited research has been conducted to address the baseline selection issues. Poor choices of baselines limit the ability of one-vs-one (1-vs-1) explanations for multi-class classifiers, which means the attribution methods were not able to explain why an input belongs to its original class but not the other specified target class. 1-vs-1 explanation is crucial when certain classes are more similar than others, e.g. two bird types among multiple animals, by focusing on key differentiating features rather than shared features across classes. In this paper, we present GAN-based Model EXplainability (GANMEX), a novel approach applying Generative Adversarial Networks (GAN) by incorporating the to-be-explained classifier as part of the adversarial networks. Our approach effectively selects the counterfactual baseline as the closest realistic sample belong to the target class, which allows attribution methods to provide true 1-vs-1 explanations. We showed that GANMEX baselines improved the saliency maps and led to stronger performance on perturbation-based evaluation metrics over the existing baselines. Existing attribution results are known for being insensitive to model randomization, and we demonstrated that GANMEX baselines led to better outcome under the cascading randomization of the model.

(2 min) In this paper, we study the bandits with knapsacks (BwK) problem and develop a primal-dual based algorithm that achieves a problem-dependent logarithmic regret bound. The BwK problem extends the multi-arm bandit (MAB) problem to model the resource consumption associated with playing each arm, and the existing BwK literature has been mainly focused on deriving asymptotically optimal distribution-free regret bounds. We first study the primal and dual linear programs underlying the BwK problem. From this primal-dual perspective, we discover symmetry between arms and knapsacks, and then propose a new notion of sub-optimality measure for the BwK problem. The sub-optimality measure highlights the important role of knapsacks in determining algorithm regret and inspires the design of our two-phase algorithm. In the first phase, the algorithm identifies the optimal arms and the binding knapsacks, and in the second phase, it exhausts the binding knapsacks via playing the optimal arms through an adaptive procedure. Our regret upper bound involves the proposed sub-optimality measure and it has a logarithmic dependence on length of horizon $T$ and a polynomial dependence on $m$ (the numbers of arms) and $d$ (the number of knapsacks). To the best of our knowledge, this is the first problem-dependent logarithmic regret bound for solving the general BwK problem.

(2 min) In Federated edge learning (FEEL), energy-constrained devices at the network edge consume significant energy when training and uploading their local machine learning models, leading to a decrease in their lifetime. This work proposes novel solutions for energy-efficient FEEL by jointly considering local training data, available computation, and communications resources, and deadline constraints of FEEL rounds to reduce energy consumption. This paper considers a system model where the edge server is equipped with multiple antennas employing beamforming techniques to communicate with the local users through orthogonal channels. Specifically, we consider a problem that aims to find the optimal user's resources, including the fine-grained selection of relevant training samples, bandwidth, transmission power, beamforming weights, and processing speed with the goal of minimizing the total energy consumption given a deadline constraint on the communication rounds of FEEL. Then, we devise tractable solutions by first proposing a novel fine-grained training algorithm that excludes less relevant training samples and effectively chooses only the samples that improve the model's performance. After that, we derive closed-form solutions, followed by a Golden-Section-based iterative algorithm to find the optimal computation and communication resources that minimize energy consumption. Experiments using MNIST and CIFAR-10 datasets demonstrate that our proposed algorithms considerably outperform the state-of-the-art solutions as energy consumption decreases by 79% for MNIST and 73% for CIFAR-10 datasets.

(2 min) Achieving a high prediction rate is a crucial task in fault detection. Although various classification procedures are available, none of them can give high accuracy in all applications. Therefore, in this paper, a novel multi-classifier fusion approach is developed to boost the performance of the individual classifiers. This is acquired by using Dempster-Shafer theory (DST). However, in cases with conflicting evidences, the DST may give counter-intuitive results. In this regard, a preprocessing technique based on a new metric is devised in order to measure and mitigate the conflict between the evidences. To evaluate and validate the effectiveness of the proposed approach, the method is applied to 15 benchmarks datasets from UCI and KEEL. Further, it is applied for classifying polycrystalline Nickel alloy first-stage turbine blades based on their broadband vibrational response. Through statistical analysis with different noise levels, and by comparing with four state-of-the-art fusion techniques, it is shown that that the proposed method improves the classification accuracy and outperforms the individual classifiers.

(2 min) We investigate the adversarial robustness of CNNs from the perspective of channel-wise activations. By comparing \textit{non-robust} (normally trained) and \textit{robustified} (adversarially trained) models, we observe that adversarial training (AT) robustifies CNNs by aligning the channel-wise activations of adversarial data with those of their natural counterparts. However, the channels that are \textit{negatively-relevant} (NR) to predictions are still over-activated when processing adversarial data. Besides, we also observe that AT does not result in similar robustness for all classes. For the robust classes, channels with larger activation magnitudes are usually more \textit{positively-relevant} (PR) to predictions, but this alignment does not hold for the non-robust classes. Given these observations, we hypothesize that suppressing NR channels and aligning PR ones with their relevances further enhances the robustness of CNNs under AT. To examine this hypothesis, we introduce a novel mechanism, i.e., \underline{C}hannel-wise \underline{I}mportance-based \underline{F}eature \underline{S}election (CIFS). The CIFS manipulates channels' activations of certain layers by generating non-negative multipliers to these channels based on their relevances to predictions. Extensive experiments on benchmark datasets including CIFAR10 and SVHN clearly verify the hypothesis and CIFS's effectiveness of robustifying CNNs.

(2 min) The move from boxed products to services and the widespread adoption of cloud computing has had a huge impact on the software development life cycle and DevOps processes. Particularly, incident management has become critical for developing and operating large-scale services. Prior work on incident management has heavily focused on the challenges with incident triaging and de-duplication. In this work, we address the fundamental problem of structured knowledge extraction from service incidents. We have built SoftNER, a framework for mining Knowledge Graphs from incident reports. First, we build a novel multi-task learning based BiLSTM-CRF model which leverages not just the semantic context but also the data-types for extracting factual information in the form of named entities. Next, we present an approach to mine relations between the named entities for automatically constructing knowledge graphs. We have deployed SoftNER at Microsoft, a major cloud service provider and have evaluated it on more than 2 months of cloud incidents. We show that the unsupervised machine learning pipeline has a high precision of 0.96. Our multi-task learning based deep learning model also outperforms the state-of-the-art NER models. Lastly, using the knowledge extracted by SoftNER, we are able to build accurate models for applications such as incident triaging and recommending entities based on their relevance to incident titles.

(2 min) Methods that sparsify a network at initialization are important in practice because they greatly improve the efficiency of both learning and inference. Our work is based on a recently proposed decomposition of the Neural Tangent Kernel (NTK) that has decoupled the dynamics of the training process into a data-dependent component and an architecture-dependent kernel - the latter referred to as Path Kernel. That work has shown how to design sparse neural networks for faster convergence, without any training data, using the Synflow-L2 algorithm. We first show that even though Synflow-L2 is optimal in terms of convergence, for a given network density, it results in sub-networks with "bottleneck" (narrow) layers - leading to poor performance as compared to other data-agnostic methods that use the same number of parameters. Then we propose a new method to construct sparse networks, without any training data, referred to as Paths with Higher-Edge Weights (PHEW). PHEW is a probabilistic network formation method based on biased random walks that only depends on the initial weights. It has similar path kernel properties as Synflow-L2 but it generates much wider layers, resulting in better generalization and performance. PHEW achieves significant improvements over the data-independent SynFlow and SynFlow-L2 methods at a wide range of network densities.

(2 min) We present a method to estimate an HDR environment map from a narrow field-of-view LDR camera image in real-time. This enables perceptually appealing reflections and shading on virtual objects of any material finish, from mirror to diffuse, rendered into a real physical environment using augmented reality. Our method is based on our efficient convolutional neural network architecture, EnvMapNet, trained end-to-end with two novel losses, ProjectionLoss for the generated image, and ClusterLoss for adversarial training. Through qualitative and quantitative comparison to state-of-the-art methods, we demonstrate that our algorithm reduces the directional error of estimated light sources by more than 50%, and achieves 3.7 times lower Frechet Inception Distance (FID). We further showcase a mobile application that is able to run our neural network model in under 9 ms on an iPhone XS, and render in real-time, visually coherent virtual objects in previously unseen real-world environments.

(2 min) This work explores the signal awareness of AI models for source code understanding. Using a software vulnerability detection use case, we evaluate the models' ability to capture the correct vulnerability signals to produce their predictions. Our prediction-preserving input minimization (P2IM) approach systematically reduces the original source code to a minimal snippet which a model needs to maintain its prediction. The model's reliance on incorrect signals is then uncovered when the vulnerability in the original code is missing in the minimal snippet, both of which the model however predicts as being vulnerable. We measure the signal awareness of models using a new metric we propose- Signal-aware Recall (SAR). We apply P2IM on three different neural network architectures across multiple datasets. The results show a sharp drop in the model's Recall from the high 90s to sub-60s with the new metric, highlighting that the models are presumably picking up a lot of noise or dataset nuances while learning their vulnerability detection logic. Although the drop in model performance may be perceived as an adversarial attack, but this isn't P2IM's objective. The idea is rather to uncover the signal-awareness of a black-box model in a data-driven manner via controlled queries. SAR's purpose is to measure the impact of task-agnostic model training, and not to suggest a shortcoming in the Recall metric. The expectation, in fact, is for SAR to match Recall in the ideal scenario where the model truly captures task-specific signals.

(2 min) Graph Neural Networks (GNNs) are limited in their expressive power, struggle with long-range interactions and lack a principled way to model higher-order structures. These problems can be attributed to the strong coupling between the computational graph and the input graph structure. The recently proposed Message Passing Simplicial Networks naturally decouple these elements by performing message passing on the clique complex of the graph. Nevertheless, these models are severely constrained by the rigid combinatorial structure of Simplicial Complexes (SCs). In this work, we extend recent theoretical results on SCs to regular Cell Complexes, topological objects that flexibly subsume SCs and graphs. We show that this generalisation provides a powerful set of graph ``lifting'' transformations, each leading to a unique hierarchical message passing procedure. The resulting methods, which we collectively call CW Networks (CWNs), are strictly more powerful than the WL test and, in certain cases, not less powerful than the 3-WL test. In particular, we demonstrate the effectiveness of one such scheme, based on rings, when applied to molecular graph problems. The proposed architecture benefits from provably larger expressivity than commonly used GNNs, principled modelling of higher-order signals and from compressing the distances between nodes. We demonstrate that our model achieves state-of-the-art results on a variety of molecular datasets.

(3 min) The goal of this work is to characterize the representational impact that foveation operations have for machine vision systems, inspired by the foveated human visual system, which has higher acuity at the center of gaze and texture-like encoding in the periphery. To do so, we introduce models consisting of a first-stage \textit{fixed} image transform followed by a second-stage \textit{learnable} convolutional neural network, and we varied the first stage component. The primary model has a foveated-textural input stage, which we compare to a model with foveated-blurred input and a model with spatially-uniform blurred input (both matched for perceptual compression), and a final reference model with minimal input-based compression. We find that: 1) the foveated-texture model shows similar scene classification accuracy as the reference model despite its compressed input, with greater i.i.d. generalization than the other models; 2) the foveated-texture model has greater sensitivity to high-spatial frequency information and greater robustness to occlusion, w.r.t the comparison models; 3) both the foveated systems, show a stronger center image-bias relative to the spatially-uniform systems even with a weight sharing constraint. Critically, these results are preserved over different classical CNN architectures throughout their learning dynamics. Altogether, this suggests that foveation with peripheral texture-based computations yields an efficient, distinct, and robust representational format of scene information, and provides symbiotic computational insight into the representational consequences that texture-based peripheral encoding may have for processing in the human visual system, while also potentially inspiring the next generation of computer vision models via spatially-adaptive computation. Code + Data available here: https://github.com/ArturoDeza/EmergentProperties

(2 min) It is challenging learning from demonstrated observation-only trajectories in a non-time-aligned environment because most imitation learning methods aim to imitate experts by following the demonstration step-by-step. However, aligned demonstrations are seldom obtainable in real-world scenarios. In this work, we propose a new imitation learning approach called Hierarchical Imitation Learning from Observation(HILONet), which adopts a hierarchical structure to choose feasible sub-goals from demonstrated observations dynamically. Our method can solve all kinds of tasks by achieving these sub-goals, whether it has a single goal position or not. We also present three different ways to increase sample efficiency in the hierarchical structure. We conduct extensive experiments using several environments. The results show the improvement in both performance and learning efficiency.

(2 min) We develop an algorithm for sequential adaptation of a classifier that is trained for a source domain to generalize in an unannotated target domain. We consider that the model has been trained on the source domain annotated data and then it needs to be adapted using the target domain unannotated data when the source domain data is not accessible. We align the distributions of the source and the target domains in a discriminative embedding space via an intermediate internal distribution. This distribution is estimated using the source data representations in the embedding. We conduct experiments on four benchmarks to demonstrate the method is effective and compares favorably against existing methods.

(2 min) This paper discusses experimental design to estimate welfare-maximizing policies. We consider a setting where units are organized into large, finitely many independent clusters and interact over unobserved dimensions within each cluster. The contribution of this paper is two-fold. First, we construct a test for whether a welfare-improving treatment configuration exists and hence worth learning by conducting a larger scale experiment. Second, we introduce an adaptive randomization procedure to estimate welfare-maximizing individual treatment allocation rules valid under unobserved interference. We derive asymptotic properties of the marginal effects estimators and finite-sample regret guarantees of the policy. Finally, we illustrate the method's advantage in simulations calibrated to an existing experiment on information diffusion.

(2 min) Multimodal VAEs seek to model the joint distribution over heterogeneous data (e.g.\ vision, language), whilst also capturing a shared representation across such modalities. Prior work has typically combined information from the modalities by reconciling idiosyncratic representations directly in the recognition model through explicit products, mixtures, or other such factorisations. Here we introduce a novel alternative, the MEME, that avoids such explicit combinations by repurposing semi-supervised VAEs to combine information between modalities implicitly through mutual supervision. This formulation naturally allows learning from partially-observed data where some modalities can be entirely missing -- something that most existing approaches either cannot handle, or do so to a limited extent. We demonstrate that MEME outperforms baselines on standard metrics across both partial and complete observation schemes on the MNIST-SVHN (image-image) and CUB (image-text) datasets. We also contrast the quality of the representations learnt by mutual supervision against standard approaches and observe interesting trends in its ability to capture relatedness between data.

(2 min) Social learning is a key component of human and animal intelligence. By taking cues from the behavior of experts in their environment, social learners can acquire sophisticated behavior and rapidly adapt to new circumstances. This paper investigates whether independent reinforcement learning (RL) agents in a multi-agent environment can learn to use social learning to improve their performance. We find that in most circumstances, vanilla model-free RL agents do not use social learning. We analyze the reasons for this deficiency, and show that by imposing constraints on the training environment and introducing a model-based auxiliary loss we are able to obtain generalized social learning policies which enable agents to: i) discover complex skills that are not learned from single-agent training, and ii) adapt online to novel environments by taking cues from experts present in the new environment. In contrast, agents trained with model-free RL or imitation learning generalize poorly and do not succeed in the transfer tasks. By mixing multi-agent and solo training, we can obtain agents that use social learning to gain skills that they can deploy when alone, even out-performing agents trained alone from the start.

(2 min) Efficient exploration in bandits is a fundamental online learning problem. We propose a variant of Thompson sampling that learns to explore better as it interacts with bandit instances drawn from an unknown prior. The algorithm meta-learns the prior and thus we call it MetaTS. We propose several efficient implementations of MetaTS and analyze it in Gaussian bandits. Our analysis shows the benefit of meta-learning and is of a broader interest, because we derive a novel prior-dependent Bayes regret bound for Thompson sampling. Our theory is complemented by empirical evaluation, which shows that MetaTS quickly adapts to the unknown prior.

(2 min) The attention module, which is a crucial component in Transformer, cannot scale efficiently to long sequences due to its quadratic complexity. Many works focus on approximating the dot-then-exponentiate softmax function in the original attention, leading to sub-quadratic or even linear-complexity Transformer architectures. However, we show that these methods cannot be applied to more powerful attention modules that go beyond the dot-then-exponentiate style, e.g., Transformers with relative positional encoding (RPE). Since in many state-of-the-art models, relative positional encoding is used as default, designing efficient Transformers that can incorporate RPE is appealing. In this paper, we propose a novel way to accelerate attention calculation for Transformers with RPE on top of the kernelized attention. Based upon the observation that relative positional encoding forms a Toeplitz matrix, we mathematically show that kernelized attention with RPE can be calculated efficiently using Fast Fourier Transform (FFT). With FFT, our method achieves $\mathcal{O}(n\log n)$ time complexity. Interestingly, we further demonstrate that properly using relative positional encoding can mitigate the training instability problem of vanilla kernelized attention. On a wide range of tasks, we empirically show that our models can be trained from scratch without any optimization issues. The learned model performs better than many efficient Transformer variants and is faster than standard Transformer in the long-sequence regime.

(2 min) We propose two algorithms that use linear function approximation (LFA) for stochastic shortest path (SSP) and bound their regret over $K$ episodes. When all stationary policies are proper, our first algorithm obtains sublinear regret ($K^{3/4}$), is computationally efficient, and uses stationary policies. This is the first LFA algorithm with these three properties, to the best of our knowledge. Our second algorithm improves the regret to $\sqrt{K}$ when the feature vectors satisfy certain assumptions. Both algorithms are special cases of a more general one, which has $\sqrt{K}$ regret for general features given access to a certain computation oracle. These algorithms and regret bounds are the first for SSP with function approximation.

(2 min) Biological systems perceive the world by simultaneously processing high-dimensional inputs from modalities as diverse as vision, audition, touch, proprioception, etc. The perception models used in deep learning on the other hand are designed for individual modalities, often relying on domain-specific assumptions such as the local grid structures exploited by virtually all existing vision models. These priors introduce helpful inductive biases, but also lock models to individual modalities. In this paper we introduce the Perceiver - a model that builds upon Transformers and hence makes few architectural assumptions about the relationship between its inputs, but that also scales to hundreds of thousands of inputs, like ConvNets. The model leverages an asymmetric attention mechanism to iteratively distill inputs into a tight latent bottleneck, allowing it to scale to handle very large inputs. We show that this architecture is competitive with or outperforms strong, specialized models on classification tasks across various modalities: images, point clouds, audio, video, and video+audio. The Perceiver obtains performance comparable to ResNet-50 and ViT on ImageNet without 2D convolutions by directly attending to 50,000 pixels. It is also competitive in all modalities in AudioSet.

(2 min) The empirical success of deep learning is often attributed to SGD's mysterious ability to avoid sharp local minima in the loss landscape, as sharp minima are known to lead to poor generalization. Recently, empirical evidence of heavy-tailed gradient noise was reported in many deep learning tasks, and it was shown in \c{S}im\c{s}ekli (2019a,b) that SGD can escape sharp local minima under the presence of such heavy-tailed gradient noise, providing a partial solution to the mystery. In this work, we analyze a popular variant of SGD where gradients are truncated above a fixed threshold. We show that it achieves a stronger notion of avoiding sharp minima: it can effectively eliminate sharp local minima entirely from its training trajectory. We characterize the dynamics of truncated SGD driven by heavy-tailed noises. First, we show that the truncation threshold and width of the attraction field dictate the order of the first exit time from the associated local minimum. Moreover, when the objective function satisfies appropriate structural conditions, we prove that as the learning rate decreases, the dynamics of heavy-tailed truncated SGD closely resemble those of a continuous-time Markov chain that never visits any sharp minima. Real data experiments on deep learning confirm our theoretical prediction that heavy-tailed SGD with gradient clipping finds a "flatter" local minima and achieves better generalization.

(2 min) Variational quantum algorithms (VQAs) promise efficient use of near-term quantum computers. However, training VQAs often requires an extensive amount of time and suffers from the barren plateau problem where the magnitude of the gradients vanishes with increasing number of qubits. Here, we show how to optimally train VQAs for learning quantum states. Parameterized quantum circuits can form Gaussian kernels, which we use to derive adaptive learning rates for gradient ascent. We introduce the generalized quantum natural gradient that features stability and optimized movement in parameter space. Both methods together outperform other optimization routines in training VQAs. Our methods also excel at numerically optimizing driving protocols for quantum control problems. The gradients of the VQA do not vanish when the fidelity between the initial state and the state to be learned is bounded from below. We identify a VQA for quantum simulation with such a constraint that thus can be trained free of barren plateaus. Finally, we propose the application of Gaussian kernels for quantum machine learning.

(2 min) We study {\em online} active learning of homogeneous halfspaces in $\mathbb{R}^d$ with adversarial noise where the overall probability of a noisy label is constrained to be at most $\nu$. Our main contribution is a Perceptron-like online active learning algorithm that runs in polynomial time, and under the conditions that the marginal distribution is isotropic log-concave and $\nu = \Omega(\epsilon)$, where $\epsilon \in (0, 1)$ is the target error rate, our algorithm PAC learns the underlying halfspace with near-optimal label complexity of $\tilde{O}\big(d \cdot polylog(\frac{1}{\epsilon})\big)$ and sample complexity of $\tilde{O}\big(\frac{d}{\epsilon} \big)$. Prior to this work, existing online algorithms designed for tolerating the adversarial noise are subject to either label complexity polynomial in $\frac{1}{\epsilon}$, or suboptimal noise tolerance, or restrictive marginal distributions. With the additional prior knowledge that the underlying halfspace is $s$-sparse, we obtain attribute-efficient label complexity of $\tilde{O}\big( s \cdot polylog(d, \frac{1}{\epsilon}) \big)$ and sample complexity of $\tilde{O}\big(\frac{s}{\epsilon} \cdot polylog(d) \big)$. As an immediate corollary, we show that under the agnostic model where no assumption is made on the noise rate $\nu$, our active learner achieves an error rate of $O(OPT) + \epsilon$ with the same running time and label and sample complexity, where $OPT$ is the best possible error rate achievable by any homogeneous halfspace.

(2 min) Tensor Networks (TN) offer a powerful framework to efficiently represent very high-dimensional objects. TN have recently shown their potential for machine learning applications and offer a unifying view of common tensor decomposition models such as Tucker, tensor train (TT) and tensor ring (TR). However, identifying the best tensor network structure from data for a given task is challenging. In this work, we leverage the TN formalism to develop a generic and efficient adaptive algorithm to jointly learn the structure and the parameters of a TN from data. Our method is based on a simple greedy approach starting from a rank one tensor and successively identifying the most promising tensor network edges for small rank increments. Our algorithm can adaptively identify TN structures with small number of parameters that effectively optimize any differentiable objective function. Experiments on tensor decomposition, tensor completion and model compression tasks demonstrate the effectiveness of the proposed algorithm. In particular, our method outperforms the state-of-the-art evolutionary topology search [Li and Sun, 2020] for tensor decomposition of images (while being orders of magnitude faster) and finds efficient tensor network structures to compress neural networks outperforming popular TT based approaches [Novikov et al., 2015].

(2 min) We present a novel approach to the problem of text style transfer. Unlike previous approaches requiring style-labeled training data, our method makes use of readily-available unlabeled text by relying on the implicit connection in style between adjacent sentences, and uses labeled data only at inference time. We adapt T5 (Raffel et al., 2020), a strong pretrained text-to-text model, to extract a style vector from text and use it to condition the decoder to perform style transfer. As our label-free training results in a style vector space encoding many facets of style, we recast transfers as "targeted restyling" vector operations that adjust specific attributes of the input while preserving others. We demonstrate that training on unlabeled Amazon reviews data results in a model that is competitive on sentiment transfer, even compared to models trained fully on labeled data. Furthermore, applying our novel method to a diverse corpus of unlabeled web text results in a single model capable of transferring along multiple dimensions of style (dialect, emotiveness, formality, politeness, sentiment) despite no additional training and using only a handful of exemplars at inference time.

(2 min) The renaissance of artificial neural networks was catalysed by the success of classification models, tagged by the community with the broader term supervised learning. The extraordinary results gave rise to a hype loaded with ambitious promises and overstatements. Soon the community realised that the success owed much to the availability of thousands of labelled examples and supervised learning went, for many, from glory to shame: Some criticised deep learning as a whole and others proclaimed that the way forward had to be alternatives to supervised learning: predictive, unsupervised, semi-supervised and, more recently, self-supervised learning. However, all these seem brand names, rather than actual categories of a theoretically grounded taxonomy. Moreover, the call to banish supervised learning was motivated by the questionable claim that humans learn with little or no supervision and are capable of robust out-of-distribution generalisation. Here, we review insights about learning and supervision in nature, revisit the notion that learning and generalisation are not possible without supervision or inductive biases and argue that we will make better progress if we just call it by its name.

(2 min) We address the problem of uncertainty calibration. While standard deep neural networks typically yield uncalibrated predictions, calibrated confidence scores that are representative of the true likelihood of a prediction can be achieved using post-hoc calibration methods. However, to date the focus of these approaches has been on in-domain calibration. Our contribution is two-fold. First, we show that existing post-hoc calibration methods yield highly over-confident predictions under domain shift. Second, we introduce a simple strategy where perturbations are applied to samples in the validation set before performing the post-hoc calibration step. In extensive experiments, we demonstrate that this perturbation step results in substantially better calibration under domain shift on a wide range of architectures and modelling tasks.

(2 min) We study the problem of clustering nodes in a dynamic graph, where the connections between nodes and nodes' cluster memberships may change over time, e.g., due to community migration. We first propose a dynamic stochastic block model that captures these changes, and a simple decay-based clustering algorithm that clusters nodes based on weighted connections between them, where the weight decreases at a fixed rate over time. This decay rate can then be interpreted as signifying the importance of including historical connection information in the clustering. However, the optimal decay rate may differ for clusters with different rates of turnover. We characterize the optimal decay rate for each cluster and propose a clustering method that achieves almost exact recovery of the true clusters. We then demonstrate the efficacy of our clustering algorithm with optimized decay rates on simulated graph data. Recurrent neural networks (RNNs), a popular algorithm for sequence learning, use a similar decay-based method, and we use this insight to propose two new RNN-GCN (graph convolutional network) architectures for semi-supervised graph clustering. We finally demonstrate that the proposed architectures perform well on real data compared to state-of-the-art graph clustering algorithms.

(3 min) In this paper, we propose a Light Gradient Boosting (LightGBM) to forecast dominant wave periods in oceanic waters. First, we use the data collected from CDIP buoys and apply various data filtering methods. The data filtering methods allow us to obtain a high-quality dataset for training and validation purposes. We then extract various wave-based features like wave heights, periods, skewness, kurtosis, etc., and atmospheric features like humidity, pressure, and air temperature for the buoys. Afterward, we train algorithms that use LightGBM and Extra Trees through a hv-block cross-validation scheme to forecast dominant wave periods for up to 30 days ahead. LightGBM has the R2 score of 0.94, 0.94, and 0.94 for 1-day ahead, 15-day ahead, and 30-day ahead prediction. Similarly, Extra Trees (ET) has an R2 score of 0.88, 0.86, and 0.85 for 1-day ahead, 15-day ahead, and 30 day ahead prediction. In case of the test dataset, LightGBM has R2 score of 0.94, 0.94, and 0.94 for 1-day ahead, 15-day ahead and 30-day ahead prediction. ET has R2 score of 0.88, 0.86, and 0.85 for 1-day ahead, 15-day ahead, and 30-day ahead prediction. A similar R2 score for both training and the test dataset suggests that the machine learning models developed in this paper are robust. Since the LightGBM algorithm outperforms ET for all the windows tested, it is taken as the final algorithm. Note that the performance of both methods does not decrease significantly as the forecast horizon increases. Likewise, the proposed method outperforms the numerical approaches included in this paper in the test dataset. For 1 day ahead prediction, the proposed algorithm has SI, Bias, CC, and RMSE of 0.09, 0.00, 0.97, and 1.78 compared to 0.268, 0.40, 0.63, and 2.18 for the European Centre for Medium-range Weather Forecasts (ECMWF) model, which outperforms all the other methods in the test dataset.

(2 min) Generative adversarial networks (GANs) synthesize realistic images from random latent vectors. Although manipulating the latent vectors controls the synthesized outputs, editing real images with GANs suffers from i) time-consuming optimization for projecting real images to the latent vectors, ii) or inaccurate embedding through an encoder. We propose StyleMapGAN: the intermediate latent space has spatial dimensions, and a spatially variant modulation replaces AdaIN. It makes the embedding through an encoder more accurate than existing optimization-based methods while maintaining the properties of GANs. Experimental results demonstrate that our method significantly outperforms state-of-the-art models in various image manipulation tasks such as local editing and image interpolation. Last but not least, conventional editing methods on GANs are still valid on our StyleMapGAN. Source code is available at https://github.com/naver-ai/StyleMapGAN.

(2 min) Visual explanation methods have an important role in the prognosis of the patients where the annotated data is limited or unavailable. There have been several attempts to use gradient-based attribution methods to localize pathology from medical scans without using segmentation labels. This research direction has been impeded by the lack of robustness and reliability. These methods are highly sensitive to the network parameters. In this study, we introduce a robust visual explanation method to address this problem for medical applications. We provide an innovative visual explanation algorithm for general purpose and as an example application, we demonstrate its effectiveness for quantifying lesions in the lungs caused by the Covid-19 with high accuracy and robustness without using dense segmentation labels. This approach overcomes the drawbacks of commonly used Grad-CAM and its extended versions. The premise behind our proposed strategy is that the information flow is minimized while ensuring the classifier prediction stays similar. Our findings indicate that the bottleneck condition provides a more stable severity estimation than the similar attribution methods.

(2 min) We study efficient PAC learning of homogeneous halfspaces in $\mathbb{R}^d$ in the presence of malicious noise of Valiant~(1985). This is a challenging noise model and only until recently has near-optimal noise tolerance bound been established under the mild condition that the unlabeled data distribution is isotropic log-concave. However, it remains unsettled how to obtain the optimal sample complexity simultaneously. In this work, we present a new analysis for the algorithm of Awasthi~et~al.~(2017) and show that it essentially achieves the near-optimal sample complexity bound of $\tilde{O}(d)$, improving the best known result of $\tilde{O}(d^2)$. Our main ingredient is a novel incorporation of a matrix Chernoff-type inequality to bound the spectrum of an empirical covariance matrix for well-behaved distributions, in conjunction with a careful exploration of the localization schemes of Awasthi~et~al.~(2017). We further extend the algorithm and analysis to the more general and stronger nasty noise model of Bshouty~et~al.~(2002), showing that it is still possible to achieve near-optimal noise tolerance and sample complexity in polynomial time.

(2 min) Recently, bound propagation based certified robust training methods have been proposed for training neural networks with certifiable robustness guarantees. Despite that state-of-the-art (SOTA) methods including interval bound propagation (IBP) and CROWN-IBP have per-batch training complexity similar to standard neural network training, they usually use a long warmup schedule with hundreds or thousands epochs to reach SOTA performance and are thus still costly. In this paper, we identify two important issues in existing methods, namely exploded bounds at initialization, and the imbalance in ReLU activation states. These two issues make certified training difficult and unstable, and thereby long warmup schedules were needed in prior works. To mitigate these issues and conduct certified training with shorter warmup, we propose three improvements: 1) We derive a new weight initialization method for IBP training; 2) We propose to fully add Batch Normalization (BN) to each layer in the model, since we find BN can reduce the imbalance in ReLU activation states; 3) We also design regularization to explicitly tighten certified bounds and balance ReLU activation states. In our experiments, we are able to obtain 65.03% verified error on CIFAR-10 ($\epsilon=\frac{8}{255}$) and 82.36% verified error on TinyImageNet ($\epsilon=\frac{1}{255}$) using very short training schedules (160 and 80 total epochs, respectively), outperforming literature SOTA trained with hundreds or thousands epochs under the same network architecture.

(2 min) Natural-gradient descent on structured parameter spaces (e.g., low-rank covariances) is computationally challenging due to complicated inverse Fisher-matrix computations. We address this issue for optimization, inference, and search problems by using \emph{local-parameter coordinates}. Our method generalizes an existing evolutionary-strategy method, recovers Newton and Riemannian-gradient methods as special cases, and also yields new tractable natural-gradient algorithms for learning flexible covariance structures of Gaussian and Wishart-based distributions via \emph{matrix groups}. We show results on a range of applications on deep learning, variational inference, and evolution strategies. Our work opens a new direction for scalable structured geometric methods via local parameterizations.

(2 min) Minimizing cross-entropy over the softmax scores of a linear map composed with a high-capacity encoder is arguably the most popular choice for training neural networks on supervised learning tasks. However, recent works show that one can directly optimize the encoder instead, to obtain equally (or even more) discriminative representations via a supervised variant of a contrastive objective. In this work, we address the question whether there are fundamental differences in the sought-for representation geometry in the output space of the encoder at minimal loss. Specifically, we prove, under mild assumptions, that both losses attain their minimum once the representations of each class collapse to the vertices of a regular simplex, inscribed in a hypersphere. We provide empirical evidence that this configuration is attained in practice and that reaching a close-to-optimal state typically indicates good generalization performance. Yet, the two losses show remarkably different optimization behavior. The number of iterations required to perfectly fit to data scales superlinearly with the amount of randomly flipped labels for the supervised contrastive loss. This is in contrast to the approximately linear scaling previously reported for networks trained with cross-entropy.

(2 min) Neural Ordinary Differential Equations (ODE) are a promising approach to learn dynamic models from time-series data in science and engineering applications. This work aims at learning Neural ODE for stiff systems, which are usually raised from chemical kinetic modeling in chemical and biological systems. We first show the challenges of learning neural ODE in the classical stiff ODE systems of Robertson's problem and propose techniques to mitigate the challenges associated with scale separations in stiff systems. We then present successful demonstrations in stiff systems of Robertson's problem and an air pollution problem. The demonstrations show that the usage of deep networks with rectified activations, proper scaling of the network outputs as well as loss functions, and stabilized gradient calculations are the key techniques enabling the learning of stiff neural ODE. The success of learning stiff neural ODE opens up possibilities of using neural ODEs in applications with widely varying time-scales, like chemical dynamics in energy conversion, environmental engineering, and the life sciences.

(2 min) Reflecting on the last few years, the biggest breakthroughs in deep reinforcement learning (RL) have been in the discrete action domain. Robotic manipulation, however, is inherently a continuous control environment, but these continuous control reinforcement learning algorithms often depend on actor-critic methods that are sample-inefficient and inherently difficult to train, due to the joint optimisation of the actor and critic. To that end, we explore how we can bring the stability of discrete action RL algorithms to the robot manipulation domain. We extend the recently released ARM algorithm, by replacing the continuous next-best pose agent with a discrete next-best pose agent. Discretisation of rotation is trivial given its bounded nature, while translation is inherently unbounded, making discretisation difficult. We formulate the translation prediction as the voxel prediction problem by discretising the 3D space; however, voxelisation of a large workspace is memory intensive and would not work with a high density of voxels, crucial to obtaining the resolution needed for robotic manipulation. We therefore propose to apply this voxel prediction in a coarse-to-fine manner by gradually increasing the resolution. In each step, we extract the highest valued voxel as the predicted location, which is then used as the centre of the higher-resolution voxelisation in the next step. This coarse-to-fine prediction is applied over several steps, giving a near-lossless prediction of the translation. We show that our new coarse-to-fine algorithm is able to accomplish RLBench tasks much more efficiently than the continuous control equivalent, and even train some real-world tasks, tabular rasa, in less than 7 minutes, with only 3 demonstrations. Moreover, we show that by moving to a voxel representation, we are able to easily incorporate observations from multiple cameras.

(2 min) We introduce feature alignment, a technique for obtaining approximate reversibility in artificial neural networks. By means of feature extraction, we can train a neural network to learn an estimated map for its reverse process from outputs to inputs. Combined with variational autoencoders, we can generate new samples from the same statistics as the training data. Improvements of the results are obtained by using concepts from generative adversarial networks. Finally, we show that the technique can be modified for training neural networks locally, saving computational memory resources. Applying these techniques, we report results for three vision generative tasks: MNIST, CIFAR-10, and celebA.

(2 min) This paper deals with the problem of localization in a cellular network in a dense urban scenario. Global Navigation Satellite Systems typically perform poorly in urban environments, where the likelihood of line-of-sight conditions between the devices and the satellites is low, and thus alternative localization methods are required for good accuracy. We present a deep learning method for localization, based merely on pathloss, which does not require any increase in computation complexity at the user devices with respect to the device standard operations, unlike methods that rely on time of arrival or angle of arrival information. In a wireless network, user devices scan the base station beacon slots and identify the few strongest base station signals for handover and user-base station association purposes. In the proposed method, the user to be localized simply reports such received signal strengths to a central processing unit, which may be located in the cloud. For each base station we have good approximation of the pathloss at every location in a dense grid in the map. This approximation is provided by RadioUNet, a deep learning-based simulator of pathloss functions in urban environment, that we have previously proposed and published. Using the estimated pathloss radio maps of all base stations and the corresponding reported signal strengths, the proposed deep learning algorithm can extract a very accurate localization of the user. The proposed method, called LocUNet, enjoys high robustness to inaccuracies in the estimated radio maps. We demonstrate this by numerical experiments, which obtain state-of-the-art results.

(2 min) Variational inference (VI) plays an essential role in approximate Bayesian inference due to its computational efficiency and broad applicability. Crucial to the performance of VI is the selection of the associated divergence measure, as VI approximates the intractable distribution by minimizing this divergence. In this paper we propose a meta-learning algorithm to learn the divergence metric suited for the task of interest, automating the design of VI methods. In addition, we learn the initialization of the variational parameters without additional cost when our method is deployed in the few-shot learning scenarios. We demonstrate our approach outperforms standard VI on Gaussian mixture distribution approximation, Bayesian neural network regression, image generation with variational autoencoders and recommender systems with a partial variational autoencoder.

(3 min) Fluorescence microscopy allows for a detailed inspection of cells, cellular networks, and anatomical landmarks by staining with a variety of carefully-selected markers visualized as color channels. Quantitative characterization of structures in acquired images often relies on automatic image analysis methods. Despite the success of deep learning methods in other vision applications, their potential for fluorescence image analysis remains underexploited. One reason lies in the considerable workload required to train accurate models, which are normally specific for a given combination of markers, and therefore applicable to a very restricted number of experimental settings. We herein propose Marker Sampling and Excite, a neural network approach with a modality sampling strategy and a novel attention module that together enable (i) flexible training with heterogeneous datasets with combinations of markers and (ii) successful utility of learned models on arbitrary subsets of markers prospectively. We show that our single neural network solution performs comparably to an upper bound scenario where an ensemble of many networks is na\"ively trained for each possible marker combination separately. In addition, we demonstrate the feasibility of this framework in high-throughput biological analysis by revising a recent quantitative characterization of bone marrow vasculature in 3D confocal microscopy datasets and further confirm the validity of our approach on an additional, significantly different dataset of microvessels in fetal liver tissues. Not only can our work substantially ameliorate the use of deep learning in fluorescence microscopy analysis, but it can also be utilized in other fields with incomplete data acquisitions and missing modalities.

(2 min) As machine learning models are increasingly used in critical decision-making settings (e.g., healthcare, finance), there has been a growing emphasis on developing methods to explain model predictions. Such \textit{explanations} are used to understand and establish trust in models and are vital components in machine learning pipelines. Though explanations are a critical piece in these systems, there is little understanding about how they are vulnerable to manipulation by adversaries. In this paper, we discuss how two broad classes of explanations are vulnerable to manipulation. We demonstrate how adversaries can design biased models that manipulate model agnostic feature attribution methods (e.g., LIME \& SHAP) and counterfactual explanations that hill-climb during the counterfactual search (e.g., Wachter's Algorithm \& DiCE) into \textit{concealing} the model's biases. These vulnerabilities allow an adversary to deploy a biased model, yet explanations will not reveal this bias, thereby deceiving stakeholders into trusting the model. We evaluate the manipulations on real world data sets, including COMPAS and Communities \& Crime, and find explanations can be manipulated in practice.

(2 min) Many users implicitly assume that software can only be exploited after it is installed. However, recent supply-chain attacks demonstrate that application integrity must be ensured during installation itself. We introduce SIGL, a new tool for detecting malicious behavior during software installation. SIGL collects traces of system call activity, building a data provenance graph that it analyzes using a novel autoencoder architecture with a graph long short-term memory network (graph LSTM) for the encoder and a standard multilayer perceptron for the decoder. SIGL flags suspicious installations as well as the specific installation-time processes that are likely to be malicious. Using a test corpus of 625 malicious installers containing real-world malware, we demonstrate that SIGL has a detection accuracy of 96%, outperforming similar systems from industry and academia by up to 87% in precision and recall and 45% in accuracy. We also demonstrate that SIGL can pinpoint the processes most likely to have triggered malicious behavior, works on different audit platforms and operating systems, and is robust to training data contamination and adversarial attack. It can be used with application-specific models, even in the presence of new software versions, as well as application-agnostic meta-models that encompass a wide range of applications and installers.

(2 min) We address the problems of identifying malware in network telemetry logs and providing \emph{indicators of compromise} -- comprehensible explanations of behavioral patterns that identify the threat. In our system, an array of specialized detectors abstracts network-flow data into comprehensible \emph{network events} in a first step. We develop a neural network that processes this sequence of events and identifies specific threats, malware families and broad categories of malware. We then use the \emph{integrated-gradients} method to highlight events that jointly constitute the characteristic behavioral pattern of the threat. We compare network architectures based on CNNs, LSTMs, and transformers, and explore the efficacy of unsupervised pre-training experimentally on large-scale telemetry data. We demonstrate how this system detects njRAT and other malware based on behavioral patterns.

(2 min) As machine learning models grow more complex and their applications become more high-stakes, tools for explaining model predictions have become increasingly important. Despite the widespread use of explainability techniques, evaluating and comparing different feature attribution methods remains challenging: evaluations ideally require human studies, and empirical evaluation metrics are often computationally prohibitive on real-world datasets. In this work, we address this issue by releasing XAI-Bench: a suite of synthetic datasets along with a library for benchmarking feature attribution algorithms. Unlike real-world datasets, synthetic datasets allow the efficient computation of conditional expected values that are needed to evaluate ground-truth Shapley values and other metrics. The synthetic datasets we release offer a wide variety of parameters that can be configured to simulate real-world data. We demonstrate the power of our library by benchmarking popular explainability techniques across several evaluation metrics and identifying failure modes for popular explainers. The efficiency of our library will help bring new explainability methods from development to deployment.

(2 min) Detecting out-of-distribution (OOD) inputs is critical for safely deploying deep learning models in an open-world setting. However, existing OOD detection solutions can be brittle in the open world, facing various types of adversarial OOD inputs. While methods leveraging auxiliary OOD data have emerged, our analysis on illuminative examples reveals a key insight that the majority of auxiliary OOD examples may not meaningfully improve or even hurt the decision boundary of the OOD detector, which is also observed in empirical results on real data. In this paper, we provide a theoretically motivated method, Adversarial Training with informative Outlier Mining (ATOM), which improves the robustness of OOD detection. We show that, by mining informative auxiliary OOD data, one can significantly improve OOD detection performance, and somewhat surprisingly, generalize to unseen adversarial attacks. ATOM achieves state-of-the-art performance under a broad family of classic and adversarial OOD evaluation tasks. For example, on the CIFAR-10 in-distribution dataset, ATOM reduces the FPR (at TPR 95%) by up to 57.99% under adversarial OOD inputs, surpassing the previous best baseline by a large margin.

(2 min) Generative Adversarial Networks are notoriously challenging to train. The underlying minmax optimization is highly susceptible to the variance of the stochastic gradient and the rotational component of the associated game vector field. To tackle these challenges, we propose the Lookahead algorithm for minmax optimization, originally developed for single objective minimization only. The backtracking step of our Lookahead-minmax naturally handles the rotational game dynamics, a property which was identified to be key for enabling gradient ascent descent methods to converge on challenging examples often analyzed in the literature. Moreover, it implicitly handles high variance without using large mini-batches, known to be essential for reaching state of the art performance. Experimental results on MNIST, SVHN, CIFAR-10, and ImageNet demonstrate a clear advantage of combining Lookahead-minmax with Adam or extragradient, in terms of performance and improved stability, for negligible memory and computational cost. Using 30-fold fewer parameters and 16-fold smaller minibatches we outperform the reported performance of the class-dependent BigGAN on CIFAR-10 by obtaining FID of 12.19 without using the class labels, bringing state-of-the-art GAN training within reach of common computational resources.

(2 min) Recently, the emergence of pre-trained models (PTMs) has brought natural language processing (NLP) to a new era. In this survey, we provide a comprehensive review of PTMs for NLP. We first briefly introduce language representation learning and its research progress. Then we systematically categorize existing PTMs based on a taxonomy with four perspectives. Next, we describe how to adapt the knowledge of PTMs to the downstream tasks. Finally, we outline some potential directions of PTMs for future research. This survey is purposed to be a hands-on guide for understanding, using, and developing PTMs for various NLP tasks.

(2 min) One of the distinct features of this century has been the population of older adults which has been on a constant rise. Elderly people have several needs and requirements due to physical disabilities, cognitive issues, weakened memory and disorganized behavior, that they face with increasing age. The extent of these limitations also differs according to the varying diversities in elderly, which include age, gender, background, experience, skills, knowledge and so on. These varying needs and challenges with increasing age, limits abilities of older adults to perform Activities of Daily Living (ADLs) in an independent manner. To add to it, the shortage of caregivers creates a looming need for technology-based services for elderly people, to assist them in performing their daily routine tasks to sustain their independent living and active aging. To address these needs, this work consists of making three major contributions in this field. First, it provides a rather comprehensive review of assisted living technologies aimed at helping elderly people to perform ADLs. Second, the work discusses the challenges identified through this review, that currently exist in the context of implementation of assisted living services for elderly care in Smart Homes and Smart Cities. Finally, the work also outlines an approach for implementation, extension and integration of the existing works in this field for development of a much-needed framework that can provide personalized assistance and user-centered behavior interventions to elderly as per their varying and ever-changing needs.

(2 min) Despite its paramount importance for manifold use cases (e.g., in the health care industry, sports, rehabilitation and fitness assessment), sufficiently valid and reliable gait parameter measurement is still limited to high-tech gait laboratories mostly. Here, we demonstrate the excellent validity and test-retest repeatability of a novel gait assessment system which is built upon modern convolutional neural networks to extract three-dimensional skeleton joints from monocular frontal-view videos of walking humans. The validity study is based on a comparison to the GAITRite pressure-sensitive walkway system. All measured gait parameters (gait speed, cadence, step length and step time) showed excellent concurrent validity for multiple walk trials at normal and fast gait speeds. The test-retest-repeatability is on the same level as the GAITRite system. In conclusion, we are convinced that our results can pave the way for cost, space and operationally effective gait analysis in broad mainstream applications. Most sensor-based systems are costly, must be operated by extensively trained personnel (e.g., motion capture systems) or - even if not quite as costly - still possess considerable complexity (e.g., wearable sensors). In contrast, a video sufficient for the assessment method presented here can be obtained by anyone, without much training, via a smartphone camera.

(2 min) Imitation learning is an effective and safe technique to train robot policies in the real world because it does not depend on an expensive random exploration process. However, due to the lack of exploration, learning policies that generalize beyond the demonstrated behaviors is still an open challenge. We present a novel imitation learning framework to enable robots to 1) learn complex real world manipulation tasks efficiently from a small number of human demonstrations, and 2) synthesize new behaviors not contained in the collected demonstrations. Our key insight is that multi-task domains often present a latent structure, where demonstrated trajectories for different tasks intersect at common regions of the state space. We present Generalization Through Imitation (GTI), a two-stage offline imitation learning algorithm that exploits this intersecting structure to train goal-directed policies that generalize to unseen start and goal state combinations. In the first stage of GTI, we train a stochastic policy that leverages trajectory intersections to have the capacity to compose behaviors from different demonstration trajectories together. In the second stage of GTI, we collect a small set of rollouts from the unconditioned stochastic policy of the first stage, and train a goal-directed agent to generalize to novel start and goal configurations. We validate GTI in both simulated domains and a challenging long-horizon robotic manipulation domain in the real world. Additional results and videos are available at https://sites.google.com/view/gti2020/ .

(2 min) Sketch drawings capture the salient information of visual concepts. Previous work has shown that neural networks are capable of producing sketches of natural objects drawn from a small number of classes. While earlier approaches focus on generation quality or retrieval, we explore properties of image representations learned by training a model to produce sketches of images. We show that this generative, class-agnostic model produces informative embeddings of images from novel examples, classes, and even novel datasets in a few-shot setting. Additionally, we find that these learned representations exhibit interesting structure and compositionality.

(2 min) The transition matrix, denoting the transition relationship from clean labels to noisy labels, is essential to build statistically consistent classifiers in label-noise learning. Existing methods for estimating the transition matrix rely heavily on estimating the noisy class posterior. However, the estimation error for noisy class posterior could be large due to the randomness of label noise, which would lead the transition matrix to be poorly estimated. Therefore, in this paper, we aim to solve this problem by exploiting the divide-and-conquer paradigm. Specifically, we introduce an intermediate class to avoid directly estimating the noisy class posterior. By this intermediate class, the original transition matrix can then be factorized into the product of two easy-to-estimate transition matrices. We term the proposed method the dual-T estimator. Both theoretical analyses and empirical results illustrate the effectiveness of the dual-T estimator for estimating transition matrices, leading to better classification performances.

(2 min) We investigate a stochastic counterpart of majority votes over finite ensembles of classifiers, and study its generalization properties. While our approach holds for arbitrary distributions, we instantiate it with Dirichlet distributions: this allows for a closed-form and differentiable expression for the expected risk, which then turns the generalization bound into a tractable training objective. The resulting stochastic majority vote learning algorithm achieves state-of-the-art accuracy and benefits from (non-vacuous) tight generalization bounds, in a series of numerical experiments when compared to competing algorithms which also minimize PAC-Bayes objectives -- both with uninformed (data-independent) and informed (data-dependent) priors.

(2 min) Spurred by tremendous success in pattern matching and prediction tasks, researchers increasingly resort to machine learning to aid original scientific discovery. Given large amounts of observational data about a system, can we uncover the rules that govern its evolution? Solving this task holds the great promise of fully understanding the causal interactions and being able to make reliable predictions about the system's behavior under interventions. We take a step towards answering this question for time-series data generated from systems of ordinary differential equations (ODEs). While the governing ODEs might not be identifiable from data alone, we show that combining simple regularization schemes with flexible neural ODEs can robustly recover the dynamics and causal structures from time-series data. Our results on a variety of (non)-linear first and second order systems as well as real data validate our method. We conclude by showing that we can also make accurate predictions under interventions on variables or the system itself.

(2 min) The goal of compressed sensing is to estimate a high dimensional vector from an underdetermined system of noisy linear equations. In analogy to classical compressed sensing, here we assume a generative model as a prior, that is, we assume the vector is represented by a deep generative model $G: \mathbb{R}^k \rightarrow \mathbb{R}^n$. Classical recovery approaches such as empirical risk minimization (ERM) are guaranteed to succeed when the measurement matrix is sub-Gaussian. However, when the measurement matrix and measurements are heavy-tailed or have outliers, recovery may fail dramatically. In this paper we propose an algorithm inspired by the Median-of-Means (MOM). Our algorithm guarantees recovery for heavy-tailed data, even in the presence of outliers. Theoretically, our results show our novel MOM-based algorithm enjoys the same sample complexity guarantees as ERM under sub-Gaussian assumptions. Our experiments validate both aspects of our claims: other algorithms are indeed fragile and fail under heavy-tailed and/or corrupted data, while our approach exhibits the predicted robustness.

(2 min) As predictive models are deployed into the real world, they must increasingly contend with strategic behavior. A growing body of work on strategic classification treats this problem as a Stackelberg game: the decision-maker "leads" in the game by deploying a model, and the strategic agents "follow" by playing their best response to the deployed model. Importantly, in this framing, the burden of learning is placed solely on the decision-maker, while the agents' best responses are implicitly treated as instantaneous. In this work, we argue that the order of play in strategic classification is fundamentally determined by the relative frequencies at which the decision-maker and the agents adapt to each other's actions. In particular, by generalizing the standard model to allow both players to learn over time, we show that a decision-maker that makes updates faster than the agents can reverse the order of play, meaning that the agents lead and the decision-maker follows. We observe in standard learning settings that such a role reversal can be desirable for both the decision-maker and the strategic agents. Finally, we show that a decision-maker with the freedom to choose their update frequency can induce learning dynamics that converge to Stackelberg equilibria with either order of play.

(2 min) Dynamical variational auto-encoders (DVAEs) are a class of deep generative models with latent variables, dedicated to time series data modeling. DVAEs can be considered as extensions of the variational autoencoder (VAE) that include the modeling of temporal dependencies between successive observed and/or latent vectors in data sequences. Previous work has shown the interest of DVAEs and their better performance over the VAE for speech signals (spectrogram) modeling. Independently, the VAE has been successfully applied to speech enhancement in noise, in an unsupervised noise-agnostic set-up that does not require the use of a parallel dataset of clean and noisy speech samples for training, but only requires clean speech signals. In this paper, we extend those works to DVAE-based single-channel unsupervised speech enhancement, hence exploiting both speech signals unsupervised representation learning and dynamics modeling. We propose an unsupervised speech enhancement algorithm based on the most general form of DVAEs, that we then adapt to three specific DVAE models to illustrate the versatility of the framework. More precisely, we combine DVAE-based speech priors with a noise model based on nonnegative matrix factorization, and we derive a variational expectation-maximization (VEM) algorithm to perform speech enhancement. Experimental results show that the proposed approach based on DVAEs outperforms its VAE counterpart and a supervised speech enhancement baseline.

(2 min) We introduce a new family of particle evolution samplers suitable for constrained domains and non-Euclidean geometries. Stein Variational Mirror Descent and Mirrored Stein Variational Gradient Descent minimize the Kullback-Leibler (KL) divergence to constrained target distributions by evolving particles in a dual space defined by a mirror map. Stein Variational Natural Gradient exploits non-Euclidean geometry to more efficiently minimize the KL divergence to unconstrained targets. We derive these samplers from a new class of mirrored Stein operators and adaptive kernels developed in this work. We demonstrate that these new samplers yield accurate approximations to distributions on the simplex, deliver valid confidence intervals in post-selection inference, and converge more rapidly than prior methods in large-scale unconstrained posterior inference. Finally, we establish the convergence of our new procedures under verifiable conditions on the target distribution.

(2 min) Machine learning algorithms train models from patterns of input data and target outputs, with the goal of predicting correct outputs for unseen test inputs. Here we demonstrate a problem of machine learning in vital application areas such as medical informatics or patent law that consists of the inclusion of measurements on which target outputs are deterministically defined in the representations of input data. This leads to perfect, but circular predictions based on a machine reconstruction of the known target definition, but fails on real-world data where the defining measurements may not or only incompletely be available. We present a circularity test that shows, for given datasets and black-box machine learning models, whether the target functional definition can be reconstructed and has been used in training. We argue that a transfer of research results to real-world applications requires to avoid circularity by separating measurements that define target outcomes from data representations in machine learning.

(2 min) Generative adversarial networks (GANs) have gained considerable attention owing to their ability to reproduce images. However, they can recreate training images faithfully despite image degradation in the form of blur, noise, and compression, generating similarly degraded images. To solve this problem, the recently proposed noise robust GAN (NR-GAN) provides a partial solution by demonstrating the ability to learn a clean image generator directly from noisy images using a two-generator model comprising image and noise generators. However, its application is limited to noise, which is relatively easy to decompose owing to its additive and reversible characteristics, and its application to irreversible image degradation, in the form of blur, compression, and combination of all, remains a challenge. To address these problems, we propose blur, noise, and compression robust GAN (BNCR-GAN) that can learn a clean image generator directly from degraded images without knowledge of degradation parameters (e.g., blur kernel types, noise amounts, or quality factor values). Inspired by NR-GAN, BNCR-GAN uses a multiple-generator model composed of image, blur-kernel, noise, and quality-factor generators. However, in contrast to NR-GAN, to address irreversible characteristics, we introduce masking architectures adjusting degradation strength values in a data-driven manner using bypasses before and after degradation. Furthermore, to suppress uncertainty caused by the combination of blur, noise, and compression, we introduce adaptive consistency losses imposing consistency between irreversible degradation processes according to the degradation strengths. We demonstrate the effectiveness of BNCR-GAN through large-scale comparative studies on CIFAR-10 and a generality analysis on FFHQ. In addition, we demonstrate the applicability of BNCR-GAN in image restoration.

(2 min) Imitation learning aims to extract knowledge from human experts' demonstrations or artificially created agents in order to replicate their behaviors. Its success has been demonstrated in areas such as video games, autonomous driving, robotic simulations and object manipulation. However, this replicating process could be problematic, such as the performance is highly dependent on the demonstration quality, and most trained agents are limited to perform well in task-specific environments. In this survey, we provide a systematic review on imitation learning. We first introduce the background knowledge from development history and preliminaries, followed by presenting different taxonomies within Imitation Learning and key milestones of the field. We then detail challenges in learning strategies and present research opportunities with learning policy from suboptimal demonstration, voice instructions and other associated optimization schemes.

(2 min) The objective of this paper is to develop predictive models to classify Brazilian legal proceedings in three possible classes of status: (i) archived proceedings, (ii) active proceedings, and (iii) suspended proceedings. This problem's resolution is intended to assist public and private institutions in managing large portfolios of legal proceedings, providing gains in scale and efficiency. In this paper, legal proceedings are made up of sequences of short texts called "motions." We combined several natural language processing (NLP) and machine learning techniques to solve the problem. Although working with Portuguese NLP, which can be challenging due to lack of resources, our approaches performed remarkably well in the classification task, achieving maximum accuracy of .93 and top average F1 Scores of .89 (macro) and .93 (weighted). Furthermore, we could extract and interpret the patterns learned by one of our models besides quantifying how those patterns relate to the classification task. The interpretability step is important among machine learning legal applications and gives us an exciting insight into how black-box models make decisions.

(2 min) This paper considers quantized distributed optimization algorithms in the parameter server framework of distributed training. We introduce the principle we call Differential Quantization (DQ) that prescribes that the past quantization errors should be compensated in such a way as to direct the descent trajectory of a quantized algorithm towards that of its unquantized counterpart. Assuming that the objective function is smooth and strongly convex, we prove that in the limit of large problem dimension, Differentially Quantized Gradient Descent (DQ-GD) attains a linear contraction factor of $\max\{\sigma_{\mathrm{GD}}, 2^{-R}\}$, where $\sigma_{\mathrm{GD}}$ is the contraction factor of unquantized gradient descent (GD). Thus at any $R\geq\log_2 1 /\sigma_{\mathrm{GD}}$ bits, the contraction factor of DQ-GD is the same as that of unquantized GD, i.e., there is no loss due to quantization. We show a converse demonstrating that no quantized gradient descent algorithm can converge faster than $\max\{\sigma_{\mathrm{GD}}, 2^{-R}\}$. In contrast, naively quantized GD where the worker directly quantizes the gradient barely attains $\sigma_{\mathrm{GD}} + 2^{-R}$. The principle of differential quantization continues to apply to gradient methods with momentum such as Nesterov's accelerated gradient descent, and Polyak's heavy ball method. For these algorithms as well, if the rate is above a certain threshold, there is no loss in contraction factor obtained by the differentially quantized algorithm compared to its unquantized counterpart, and furthermore, the differentially quantized heavy ball method attains the optimal contraction achievable among all (even unquantized) gradient methods. Experimental results on both simulated and real-world least-squares problems validate our theoretical analysis.

(2 min) An innovations sequence of a time series is a sequence of independent and identically distributed random variables with which the original time series has a causal representation. The innovation at a time is statistically independent of the prior history of the time series. As such, it represents the new information contained at present but not in the past. Because of its simple probability structure, an innovations sequence is the most efficient signature of the original. Unlike the principle or independent analysis (PCA/ICA) representations, an innovations sequence preserves not only the complete statistical properties but also the temporal order of the original time series. An long-standing open problem is to find a computationally tractable way to extract an innovations sequence of non-Gaussian processes. This paper presents a deep learning approach, referred to as Innovations Autoencoder (IAE), that extracts innovations sequences using a causal convolutional neural network. An application of IAE to nonparametric anomaly detection with unknown anomaly and anomaly-free models is also presented.

(2 min) Data subset selection from a large number of training instances has been a successful approach toward efficient and cost-effective machine learning. However, models trained on a smaller subset may show poor generalization ability. In this paper, our goal is to design an algorithm for selecting a subset of the training data, so that the model can be trained quickly, without significantly sacrificing on accuracy. More specifically, we focus on data subset selection for L2 regularized regression problems and provide a novel problem formulation which seeks to minimize the training loss with respect to both the trainable parameters and the subset of training data, subject to error bounds on the validation set. We tackle this problem using several technical innovations. First, we represent this problem with simplified constraints using the dual of the original training problem and show that the objective of this new representation is a monotone and alpha-submodular function, for a wide variety of modeling choices. Such properties lead us to develop SELCON, an efficient majorization-minimization algorithm for data subset selection, that admits an approximation guarantee even when the training provides an imperfect estimate of the trained model. Finally, our experiments on several datasets show that SELCON trades off accuracy and efficiency more effectively than the current state-of-the-art.

(2 min) Deep neural networks, while generalize well, are known to be sensitive to small adversarial perturbations. This phenomenon poses severe security threat and calls for in-depth investigation of the robustness of deep learning models. With the emergence of neural networks for graph structured data, similar investigations are urged to understand their robustness. It has been found that adversarially perturbing the graph structure and/or node features may result in a significant degradation of the model performance. In this work, we show from a different angle that such fragility similarly occurs if the graph contains a few bad-actor nodes, which compromise a trained graph neural network through flipping the connections to any targeted victim. Worse, the bad actors found for one graph model severely compromise other models as well. We call the bad actors ``anchor nodes'' and propose an algorithm, named GUA, to identify them. Thorough empirical investigations suggest an interesting finding that the anchor nodes often belong to the same class; and they also corroborate the intuitive trade-off between the number of anchor nodes and the attack success rate. For the dataset Cora which contains 2708 nodes, as few as six anchor nodes will result in an attack success rate higher than 80\% for GCN and other three models.

(2 min) Binarized Neural Networks (BNNs) have the potential to revolutionize the way that deep learning is carried out in edge computing platforms. However, the effectiveness of interpretability methods on these networks has not been assessed. In this paper, we compare the performance of several widely used saliency map-based interpretabilty techniques (Gradient, SmoothGrad and GradCAM), when applied to Binarized or Full Precision Neural Networks (FPNNs). We found that the basic Gradient method produces very similar-looking maps for both types of network. However, SmoothGrad produces significantly noisier maps for BNNs. GradCAM also produces saliency maps which differ between network types, with some of the BNNs having seemingly nonsensical explanations. We comment on possible reasons for these differences in explanations and present it as an example of why interpretability techniques should be tested on a wider range of network types.

(2 min) Diabetic Retinopathy (DR) is among the worlds leading vision loss causes in diabetic patients. DR is a microvascular disease that affects the eye retina, which causes vessel blockage and therefore cuts the main source of nutrition for the retina tissues. Treatment for this visual disorder is most effective when it is detected in its earliest stages, as severe DR can result in irreversible blindness. Nonetheless, DR identification requires the expertise of Ophthalmologists which is often expensive and time-consuming. Therefore, automatic detection systems were introduced aiming to facilitate the identification process, making it available globally in a time and cost-efficient manner. However, due to the limited reliable datasets and medical records for this particular eye disease, the obtained predictions accuracies were relatively unsatisfying for eye specialists to rely on them as diagnostic systems. Thus, we explored an ensemble-based learning strategy, merging a substantial selection of well-known classification algorithms in one sophisticated diagnostic model. The proposed framework achieved the highest accuracy rates among all other common classification algorithms in the area. 4 subdatasets were generated to contain the top 5 and top 10 features of the Messidor dataset, selected by InfoGainEval. and WrapperSubsetEval., accuracies of 70.7% and 75.1% were achieved on the InfoGainEval. top 5 and original dataset respectively. The results imply the impressive performance of the subdataset, which significantly conduces to a less complex classification process

(2 min) Left ventricular hypertrophy (LVH) results from chronic remodeling caused by a broad range of systemic and cardiovascular disease including hypertension, aortic stenosis, hypertrophic cardiomyopathy, and cardiac amyloidosis. Early detection and characterization of LVH can significantly impact patient care but is limited by under-recognition of hypertrophy, measurement error and variability, and difficulty differentiating etiologies of LVH. To overcome this challenge, we present EchoNet-LVH - a deep learning workflow that automatically quantifies ventricular hypertrophy with precision equal to human experts and predicts etiology of LVH. Trained on 28,201 echocardiogram videos, our model accurately measures intraventricular wall thickness (mean absolute error [MAE] 1.4mm, 95% CI 1.2-1.5mm), left ventricular diameter (MAE 2.4mm, 95% CI 2.2-2.6mm), and posterior wall thickness (MAE 1.2mm, 95% CI 1.1-1.3mm) and classifies cardiac amyloidosis (area under the curve of 0.83) and hypertrophic cardiomyopathy (AUC 0.98) from other etiologies of LVH. In external datasets from independent domestic and international healthcare systems, EchoNet-LVH accurately quantified ventricular parameters (R2 of 0.96 and 0.90 respectively) and detected cardiac amyloidosis (AUC 0.79) and hypertrophic cardiomyopathy (AUC 0.89) on the domestic external validation site. Leveraging measurements across multiple heart beats, our model can more accurately identify subtle changes in LV geometry and its causal etiologies. Compared to human experts, EchoNet-LVH is fully automated, allowing for reproducible, precise measurements, and lays the foundation for precision diagnosis of cardiac hypertrophy. As a resource to promote further innovation, we also make publicly available a large dataset of 23,212 annotated echocardiogram videos.

(2 min) We introduce a conceptually simple yet effective model for self-supervised representation learning with graph data. It follows the previous methods that generate two views of an input graph through data augmentation. However, unlike contrastive methods that focus on instance-level discrimination, we optimize an innovative feature-level objective inspired by classical Canonical Correlation Analysis. Compared with other works, our approach requires none of the parameterized mutual information estimator, additional projector, asymmetric structures, and most importantly, negative samples which can be costly. We show that the new objective essentially 1) aims at discarding augmentation-variant information by learning invariant representations, and 2) can prevent degenerated solutions by decorrelating features in different dimensions. Our theoretical analysis further provides an understanding for the new objective which can be equivalently seen as an instantiation of the Information Bottleneck Principle under the self-supervised setting. Despite its simplicity, our method performs competitively on seven public graph datasets.

(2 min) Applying artificial neural networks (ANN) to specific tasks, researchers, programmers, and other specialists usually overshot the number of convolutional layers in their designs. By implication, these ANNs hold too many parameters, which needed unnecessarily trained without impacting the result. The features, a convolutional layer can process, are strictly limited by its receptive field. By layer-wise analyzing the expansion of the receptive fields, we can reliably predict sequences of layers that will not contribute qualitatively to the inference in thegiven ANN architecture. Based on these analyses, we propose design strategies to resolve these inefficiencies, optimizing the explainability and the computational performance of ANNs. Since neither the strategies nor the analysis requires training of the actual model, these insights allow for a very efficient design process of ANNs architectures which might be automated in the future.

(2 min) Self-training based unsupervised domain adaptation (UDA) has shown great potential to address the problem of domain shift, when applying a trained deep learning model in a source domain to unlabeled target domains. However, while the self-training UDA has demonstrated its effectiveness on discriminative tasks, such as classification and segmentation, via the reliable pseudo-label selection based on the softmax discrete histogram, the self-training UDA for generative tasks, such as image synthesis, is not fully investigated. In this work, we propose a novel generative self-training (GST) UDA framework with continuous value prediction and regression objective for cross-domain image synthesis. Specifically, we propose to filter the pseudo-label with an uncertainty mask, and quantify the predictive confidence of generated images with practical variational Bayes learning. The fast test-time adaptation is achieved by a round-based alternative optimization scheme. We validated our framework on the tagged-to-cine magnetic resonance imaging (MRI) synthesis problem, where datasets in the source and target domains were acquired from different scanners or centers. Extensive validations were carried out to verify our framework against popular adversarial training UDA methods. Results show that our GST, with tagged MRI of test subjects in new target domains, improved the synthesis quality by a large margin, compared with the adversarial training UDA methods.

(2 min) Knowledge is acquired by humans through experience, and no boundary is set between the kinds of knowledge or skill levels we can achieve on different tasks at the same time. When it comes to Neural Networks, that is not the case, the major breakthroughs in the field are extremely task and domain specific. Vision and language are dealt with in separate manners, using separate methods and different datasets. In this work, we propose to use knowledge acquired by benchmark Vision Models which are trained on ImageNet to help a much smaller architecture learn to classify text. After transforming the textual data contained in the IMDB dataset to gray scale images. An analysis of different domains and the Transfer Learning method is carried out. Despite the challenge posed by the very different datasets, promising results are achieved. The main contribution of this work is a novel approach which links large pretrained models on both language and vision to achieve state-of-the-art results in different sub-fields from the original task. Without needing high compute capacity resources. Specifically, Sentiment Analysis is achieved after transferring knowledge between vision and language models. BERT embeddings are transformed into grayscale images, these images are then used as training examples for pretrained vision models such as VGG16 and ResNet Index Terms: Natural language, Vision, BERT, Transfer Learning, CNN, Domain Adaptation.

(2 min) Transfer learning has become a common solution to address training data scarcity in practice. It trains a specified student model by reusing or fine-tuning early layers of a well-trained teacher model that is usually publicly available. However, besides utility improvement, the transferred public knowledge also brings potential threats to model confidentiality, and even further raises other security and privacy issues. In this paper, we present the first comprehensive investigation of the teacher model exposure threat in the transfer learning context, aiming to gain a deeper insight into the tension between public knowledge and model confidentiality. To this end, we propose a teacher model fingerprinting attack to infer the origin of a student model, i.e., the teacher model it transfers from. Specifically, we propose a novel optimization-based method to carefully generate queries to probe the student model to realize our attack. Unlike existing model reverse engineering approaches, our proposed fingerprinting method neither relies on fine-grained model outputs, e.g., posteriors, nor auxiliary information of the model architecture or training dataset. We systematically evaluate the effectiveness of our proposed attack. The empirical results demonstrate that our attack can accurately identify the model origin with few probing queries. Moreover, we show that the proposed attack can serve as a stepping stone to facilitating other attacks against machine learning models, such as model stealing.

(2 min) Advances in neural architecture search, as well as explainability and interpretability of connectionist architectures, have been reported in the recent literature. However, our understanding of how to design Bayesian Deep Learning (BDL) hyperparameters, specifically, the depth, width and ensemble size, for robust function mapping with uncertainty quantification, is still emerging. This paper attempts to further our understanding by mapping Bayesian connectionist representations to polynomials of different orders with varying noise types and ratios. We examine the noise-contaminated polynomials to search for the combination of hyperparameters that can extract the underlying polynomial signals while quantifying uncertainties based on the noise attributes. Specifically, we attempt to study the question that an appropriate neural architecture and ensemble configuration can be found to detect a signal of any n-th order polynomial contaminated with noise having different distributions and signal-to-noise (SNR) ratios and varying noise attributes. Our results suggest the possible existence of an optimal network depth as well as an optimal number of ensembles for prediction skills and uncertainty quantification, respectively. However, optimality is not discernible for width, even though the performance gain reduces with increasing width at high values of width. Our experiments and insights can be directional to understand theoretical properties of BDL representations and to design practical solutions.

(2 min) Unsupervised domain adaptation (UDA) aims to transfer knowledge learned from a labeled source domain to an unlabeled and unseen target domain, which is usually trained on data from both domains. Access to the source domain data at the adaptation stage, however, is often limited, due to data storage or privacy issues. To alleviate this, in this work, we target source free UDA for segmentation, and propose to adapt an ``off-the-shelf" segmentation model pre-trained in the source domain to the target domain, with an adaptive batch-wise normalization statistics adaptation framework. Specifically, the domain-specific low-order batch statistics, i.e., mean and variance, are gradually adapted with an exponential momentum decay scheme, while the consistency of domain shareable high-order batch statistics, i.e., scaling and shifting parameters, is explicitly enforced by our optimization objective. The transferability of each channel is adaptively measured first from which to balance the contribution of each channel. Moreover, the proposed source free UDA framework is orthogonal to unsupervised learning methods, e.g., self-entropy minimization, which can thus be simply added on top of our framework. Extensive experiments on the BraTS 2018 database show that our source free UDA framework outperformed existing source-relaxed UDA methods for the cross-subtype UDA segmentation task and yielded comparable results for the cross-modality UDA segmentation task, compared with a supervised UDA methods with the source data.

(2 min) Intelligent systems are transforming the world, as well as our healthcare system. We propose a deep learning-based cough sound classification model that can distinguish between children with healthy versus pathological coughs such as asthma, upper respiratory tract infection (URTI), and lower respiratory tract infection (LRTI). In order to train a deep neural network model, we collected a new dataset of cough sounds, labelled with clinician's diagnosis. The chosen model is a bidirectional long-short term memory network (BiLSTM) based on Mel Frequency Cepstral Coefficients (MFCCs) features. The resulting trained model when trained for classifying two classes of coughs -- healthy or pathology (in general or belonging to a specific respiratory pathology), reaches accuracy exceeding 84\% when classifying cough to the label provided by the physicians' diagnosis. In order to classify subject's respiratory pathology condition, results of multiple cough epochs per subject were combined. The resulting prediction accuracy exceeds 91\% for all three respiratory pathologies. However, when the model is trained to classify and discriminate among the four classes of coughs, overall accuracy dropped: one class of pathological coughs are often misclassified as other. However, if one consider the healthy cough classified as healthy and pathological cough classified to have some kind of pathologies, then the overall accuracy of four class model is above 84\%. A longitudinal study of MFCC feature space when comparing pathological and recovered coughs collected from the same subjects revealed the fact that pathological cough irrespective of the underlying conditions occupy the same feature space making it harder to differentiate only using MFCC features.

(2 min) Compared with avid research activities of deep convolutional neural networks (DCNNs) in practice, the study of theoretical behaviors of DCNNs lags heavily behind. In particular, the universal consistency of DCNNs remains open. In this paper, we prove that implementing empirical risk minimization on DCNNs with expansive convolution (with zero-padding) is strongly universally consistent. Motivated by the universal consistency, we conduct a series of experiments to show that without any fully connected layers, DCNNs with expansive convolution perform not worse than the widely used deep neural networks with hybrid structure containing contracting (without zero-padding) convolution layers and several fully connected layers.

(2 min) In today's academic publishing model, especially in Computer Science, conferences commonly constitute the main platforms for releasing the latest peer-reviewed advancements in their respective fields. However, choosing a suitable academic venue for publishing one's research can represent a challenging task considering the plethora of available conferences, particularly for those at the start of their academic careers, or for those seeking to publish outside of their usual domain. In this paper, we propose GraphConfRec, a conference recommender system which combines SciGraph and graph neural networks, to infer suggestions based not only on title and abstract, but also on co-authorship and citation relationships. GraphConfRec achieves a recall@10 of up to 0.580 and a MAP of up to 0.336 with a graph attention network-based recommendation model. A user study with 25 subjects supports the positive results.

(2 min) Recently, the study on learned iterative shrinkage thresholding algorithm (LISTA) has attracted increasing attentions. A large number of experiments as well as some theories have proved the high efficiency of LISTA for solving sparse coding problems. However, existing LISTA methods are all serial connection. To address this issue, we propose a novel extragradient based LISTA (ELISTA), which has a residual structure and theoretical guarantees. In particular, our algorithm can also provide the interpretability for Res-Net to a certain extent. From a theoretical perspective, we prove that our method attains linear convergence. In practice, extensive empirical results verify the advantages of our method.

(2 min) This paper introduces a neural network approach for fitting the Lee-Carter and the Poisson Lee-Carter model on multiple populations. We develop some neural networks that replicate the structure of the individual LC models and allow their joint fitting by analysing the mortality data of all the considered populations simultaneously. The neural network architecture is specifically designed to calibrate each individual model using all available information instead of using a population-specific subset of data as in the traditional estimation schemes. A large set of numerical experiments performed on all the countries of the Human Mortality Database (HMD) shows the effectiveness of our approach. In particular, the resulting parameter estimates appear smooth and less sensitive to the random fluctuations often present in the mortality rates' data, especially for low-population countries. In addition, the forecasting performance results significantly improved as well.

(2 min) In many sequential decision-making problems (e.g., robotics control, game playing, sequential prediction), human or expert data is available containing useful information about the task. However, imitation learning (IL) from a small amount of expert data can be challenging in high-dimensional environments with complex dynamics. Behavioral cloning is a simple method that is widely used due to its simplicity of implementation and stable convergence but doesn't utilize any information involving the environment's dynamics. Many existing methods that exploit dynamics information are difficult to train in practice due to an adversarial optimization process over reward and policy approximators or biased, high variance gradient estimators. We introduce a method for dynamics-aware IL which avoids adversarial training by learning a single Q-function, implicitly representing both reward and policy. On standard benchmarks, the implicitly learned rewards show a high positive correlation with the ground-truth rewards, illustrating our method can also be used for inverse reinforcement learning (IRL). Our method, Inverse soft-Q learning (IQ-Learn) obtains state-of-the-art results in offline and online imitation learning settings, surpassing existing methods both in the number of required environment interactions and scalability in high-dimensional spaces.

(2 min) While representation learning has yielded a great success on many graph learning tasks, there is little understanding behind the structures that are being captured by these embeddings. For example, we wonder if the topological features, such as the Triangle Count, the Degree of the node and other centrality measures are concretely encoded in the embeddings. Furthermore, we ask if the presence of these structures in the embeddings is necessary for a better performance on the downstream tasks, such as clustering and classification. To address these questions, we conduct an extensive empirical study over three classes of unsupervised graph embedding models and seven different variants of Graph Autoencoders. Our results show that five topological features: the Degree, the Local Clustering Score, the Betweenness Centrality, the Eigenvector Centrality, and Triangle Count are concretely preserved in the first layer of the graph autoencoder that employs the SUM aggregation rule, under the condition that the model preserves the second-order proximity. We supplement further evidence for the presence of these features by revealing a hierarchy in the distribution of the topological features in the embeddings of the aforementioned model. We also show that a model with such properties can outperform other models on certain downstream tasks, especially when the preserved features are relevant to the task at hand. Finally, we evaluate the suitability of our findings through a test case study related to social influence prediction.

(2 min) Multi-layer feedforward networks have been used to approximate a wide range of nonlinear functions. An important and fundamental problem is to understand the learnability of a network model through its statistical risk, or the expected prediction error on future data. To the best of our knowledge, the rate of convergence of neural networks shown by existing works is bounded by at most the order of $n^{-1/4}$ for a sample size of $n$. In this paper, we show that a class of variation-constrained neural networks, with arbitrary width, can achieve near-parametric rate $n^{-1/2+\delta}$ for an arbitrarily small positive constant $\delta$. It is equivalent to $n^{-1 +2\delta}$ under the mean squared error. This rate is also observed by numerical experiments. The result indicates that the neural function space needed for approximating smooth functions may not be as large as what is often perceived. Our result also provides insight to the phenomena that deep neural networks do not easily suffer from overfitting when the number of neurons and learning parameters rapidly grow with $n$ or even surpass $n$. We also discuss the rate of convergence regarding other network parameters, including the input dimension, network layer, and coefficient norm.

(2 min) We consider stochastic convex optimization problems, where several machines act asynchronously in parallel while sharing a common memory. We propose a robust training method for the constrained setting and derive non asymptotic convergence guarantees that do not depend on prior knowledge of update delays, objective smoothness, and gradient variance. Conversely, existing methods for this setting crucially rely on this prior knowledge, which render them unsuitable for essentially all shared-resources computational environments, such as clouds and data centers. Concretely, existing approaches are unable to accommodate changes in the delays which result from dynamic allocation of the machines, while our method implicitly adapts to such changes.

(2 min) Shift workers who are essential contributors to our society, face high risks of poor health and wellbeing. To help with their problems, we collected and analyzed physiological and behavioral wearable sensor data from shift working nurses and doctors, as well as their behavioral questionnaire data and their self-reported daily health and wellbeing labels, including alertness, happiness, energy, health, and stress. We found the similarities and differences between the responses of nurses and doctors. According to the differences in self-reported health and wellbeing labels between nurses and doctors, and the correlations among their labels, we proposed a job-role based multitask and multilabel deep learning model, where we modeled physiological and behavioral data for nurses and doctors simultaneously to predict participants' next day's multidimensional self-reported health and wellbeing status. Our model showed significantly better performances than baseline models and previous state-of-the-art models in the evaluations of binary/3-class classification and regression prediction tasks. We also found features related to heart rate, sleep, and work shift contributed to shift workers' health and wellbeing.

(2 min) Recurrent Neural Networks (RNNs), more specifically their Long Short-Term Memory (LSTM) variants, have been widely used as a deep learning tool for tackling sequence-based learning tasks in text and speech. Training of such LSTM applications is computationally intensive due to the recurrent nature of hidden state computation that repeats for each time step. While sparsity in Deep Neural Nets has been widely seen as an opportunity for reducing computation time in both training and inference phases, the usage of non-ReLU activation in LSTM RNNs renders the opportunities for such dynamic sparsity associated with neuron activation and gradient values to be limited or non-existent. In this work, we identify dropout induced sparsity for LSTMs as a suitable mode of computation reduction. Dropout is a widely used regularization mechanism, which randomly drops computed neuron values during each iteration of training. We propose to structure dropout patterns, by dropping out the same set of physical neurons within a batch, resulting in column (row) level hidden state sparsity, which are well amenable to computation reduction at run-time in general-purpose SIMD hardware as well as systolic arrays. We conduct our experiments for three representative NLP tasks: language modelling on the PTB dataset, OpenNMT based machine translation using the IWSLT De-En and En-Vi datasets, and named entity recognition sequence labelling using the CoNLL-2003 shared task. We demonstrate that our proposed approach can be used to translate dropout-based computation reduction into reduced training time, with improvement ranging from 1.23x to 1.64x, without sacrificing the target metric.

(2 min) Federated Learning (FL) has become an active and promising distributed machine learning paradigm. As a result of statistical heterogeneity, recent studies clearly show that the performance of popular FL methods (e.g., FedAvg) deteriorates dramatically due to the client drift caused by local updates. This paper proposes a novel Federated Learning algorithm (called IGFL), which leverages both Individual and Group behaviors to mimic distribution, thereby improving the ability to deal with heterogeneity. Unlike existing FL methods, our IGFL can be applied to both client and server optimization. As a by-product, we propose a new attention-based federated learning in the server optimization of IGFL. To the best of our knowledge, this is the first time to incorporate attention mechanisms into federated optimization. We conduct extensive experiments and show that IGFL can significantly improve the performance of existing federated learning methods. Especially when the distributions of data among individuals are diverse, IGFL can improve the classification accuracy by about 13% compared with prior baselines.

(2 min) An increasingly common setting in machine learning involves multiple parties, each with their own data, who want to jointly make predictions on future test points. Agents wish to benefit from the collective expertise of the full set of agents to make better predictions than they would individually, but may not be willing to release their data or model parameters. In this work, we explore a decentralized mechanism to make collective predictions at test time, leveraging each agent's pre-trained model without relying on external validation, model retraining, or data pooling. Our approach takes inspiration from the literature in social science on human consensus-making. We analyze our mechanism theoretically, showing that it converges to inverse meansquared-error (MSE) weighting in the large-sample limit. To compute error bars on the collective predictions we propose a decentralized Jackknife procedure that evaluates the sensitivity of our mechanism to a single agent's prediction. Empirically, we demonstrate that our scheme effectively combines models with differing quality across the input space. The proposed consensus prediction achieves significant gains over classical model averaging, and even outperforms weighted averaging schemes that have access to additional validation data.

(2 min) Recent advances in scanning tunneling and transmission electron microscopies (STM and STEM) have allowed routine generation of large volumes of imaging data containing information on the structure and functionality of materials. The experimental data sets contain signatures of long-range phenomena such as physical order parameter fields, polarization and strain gradients in STEM, or standing electronic waves and carrier-mediated exchange interactions in STM, all superimposed onto scanning system distortions and gradual changes of contrast due to drift and/or mis-tilt effects. Correspondingly, while the human eye can readily identify certain patterns in the images such as lattice periodicities, repeating structural elements, or microstructures, their automatic extraction and classification are highly non-trivial and universal pathways to accomplish such analyses are absent. We pose that the most distinctive elements of the patterns observed in STM and (S)TEM images are similarity and (almost-) periodicity, behaviors stemming directly from the parsimony of elementary atomic structures, superimposed on the gradual changes reflective of order parameter distributions. However, the discovery of these elements via global Fourier methods is non-trivial due to variability and lack of ideal discrete translation symmetry. To address this problem, we develop shift-invariant variational autoencoders (shift-VAE) that allow disentangling characteristic repeating features in the images, their variations, and shifts inevitable for random sampling of image space. Shift-VAEs balance the uncertainty in the position of the object of interest with the uncertainty in shape reconstruction. This approach is illustrated for model 1D data, and further extended to synthetic and experimental STM and STEM 2D data.

(2 min) The increasing computational requirements of deep neural networks (DNNs) have led to significant interest in obtaining DNN models that are sparse, yet accurate. Recent work has investigated the even harder case of sparse training, where the DNN weights are, for as much as possible, already sparse to reduce computational costs during training. Existing sparse training methods are mainly empirical and often have lower accuracy relative to the dense baseline. In this paper, we present a general approach called Alternating Compressed/DeCompressed (AC/DC) training of DNNs, demonstrate convergence for a variant of the algorithm, and show that AC/DC outperforms existing sparse training methods in accuracy at similar computational budgets; at high sparsity levels, AC/DC even outperforms existing methods that rely on accurate pre-trained dense models. An important property of AC/DC is that it allows co-training of dense and sparse models, yielding accurate sparse-dense model pairs at the end of the training process. This is useful in practice, where compressed variants may be desirable for deployment in resource-constrained settings without re-doing the entire training flow, and also provides us with insights into the accuracy gap between dense and compressed models.

(2 min) Shapley values has established itself as one of the most appropriate and theoretically sound frameworks for explaining predictions from complex machine learning models. The popularity of Shapley values in the explanation setting is probably due to its unique theoretical properties. The main drawback with Shapley values, however, is that its computational complexity grows exponentially in the number of input features, making it unfeasible in many real world situations where there could be hundreds or thousands of features. Furthermore, with many (dependent) features, presenting/visualizing and interpreting the computed Shapley values also becomes challenging. The present paper introduces groupShapley: a conceptually simple approach for dealing with the aforementioned bottlenecks. The idea is to group the features, for example by type or dependence, and then compute and present Shapley values for these groups instead of for all individual features. Reducing hundreds or thousands of features to half a dozen or so, makes precise computations practically feasible and the presentation and knowledge extraction greatly simplified. We prove that under certain conditions, groupShapley is equivalent to summing the feature-wise Shapley values within each feature group. Moreover, we provide a simulation study exemplifying the differences when these conditions are not met. We illustrate the usability of the approach in a real world car insurance example, where groupShapley is used to provide simple and intuitive explanations.

(2 min) Frequently, population studies feature pyramidally-organized data represented using Hierarchical Bayesian Models (HBM) enriched with plates. These models can become prohibitively large in settings such as neuroimaging, where a sample is composed of a functional MRI signal measured on 64 thousand brain locations, across 4 measurement sessions, and at least tens of subjects. Even a reduced example on a specific cortical region of 300 brain locations features around 1 million parameters, hampering the usage of modern density estimation techniques such as Simulation-Based Inference (SBI). To infer parameter posterior distributions in this challenging class of problems, we designed a novel methodology that automatically produces a variational family dual to a target HBM. This variatonal family, represented as a neural network, consists in the combination of an attention-based hierarchical encoder feeding summary statistics to a set of normalizing flows. Our automatically-derived neural network exploits exchangeability in the plate-enriched HBM and factorizes its parameter space. The resulting architecture reduces by orders of magnitude its parameterization with respect to that of a typical SBI representation, while maintaining expressivity. Our method performs inference on the specified HBM in an amortized setup: once trained, it can readily be applied to a new data sample to compute the parameters' full posterior. We demonstrate the capability of our method on simulated data, as well as a challenging high-dimensional brain parcellation experiment. We also open up several questions that lie at the intersection between SBI techniques and structured Variational Inference.

(2 min) We study data clustering problems with $\ell_p$-norm objectives (e.g. $k$-Median and $k$-Means) in the context of individual fairness. The dataset consists of $n$ points, and we want to find $k$ centers such that (a) the objective is minimized, while (b) respecting the individual fairness constraint that every point $v$ has a center within a distance at most $r(v)$, where $r(v)$ is $v$'s distance to its $(n/k)$th nearest point. Jung, Kannan, and Lutz [FORC 2020] introduced this concept and designed a clustering algorithm with provable (approximate) fairness and objective guarantees for the $\ell_\infty$ or $k$-Center objective. Mahabadi and Vakilian [ICML 2020] revisited this problem to give a local-search algorithm for all $\ell_p$-norms. Empirically, their algorithms outperform Jung et. al.'s by a large margin in terms of cost (for $k$-Median and $k$-Means), but they incur a reasonable loss in fairness. In this paper, our main contribution is to use Linear Programming (LP) techniques to obtain better algorithms for this problem, both in theory and in practice. We prove that by modifying known LP rounding techniques, one gets a worst-case guarantee on the objective which is much better than in MV20, and empirically, this objective is extremely close to the optimal. Furthermore, our theoretical fairness guarantees are comparable with MV20 in theory, and empirically, we obtain noticeably fairer solutions. Although solving the LP {\em exactly} might be prohibitive, we demonstrate that in practice, a simple sparsification technique drastically improves the run-time of our algorithm.

(2 min) In this paper, we propose a novel conservative formulation for solving kinetic equations via neural networks. More precisely, we formulate the learning problem as a constrained optimization problem with constraints that represent the physical conservation laws. The constraints are relaxed toward the residual loss function by the Lagrangian duality. By imposing physical conservation properties of the solution as constraints of the learning problem, we demonstrate far more accurate approximations of the solutions in terms of errors and the conservation laws, for the kinetic Fokker-Planck equation and the homogeneous Boltzmann equation.

(2 min) Automatic processing of bibliographic data becomes very important in digital libraries, data science and machine learning due to its importance in keeping pace with the significant increase of published papers every year from one side and to the inherent challenges from the other side. This processing has several aspects including but not limited to I) Automatic extraction of references from PDF documents, II) Building an accurate citation graph, III) Author name disambiguation, etc. Bibliographic data is heterogeneous by nature and occurs in both structured (e.g. citation graph) and unstructured (e.g. publications) formats. Therefore, it requires data science and machine learning techniques to be processed and analysed. Here we introduce BiblioDAP'21: The 1st Workshop on Bibliographic Data Analysis and Processing.

(2 min) Accurate prediction of pedestrian and bicyclist paths is integral to the development of reliable autonomous vehicles in dense urban environments. The interactions between vehicle and pedestrian or bicyclist have a significant impact on the trajectories of traffic participants e.g. stopping or turning to avoid collisions. Although recent datasets and trajectory prediction approaches have fostered the development of autonomous vehicles yet the amount of vehicle-pedestrian (bicyclist) interactions modeled are sparse. In this work, we propose Euro-PVI, a dataset of pedestrian and bicyclist trajectories. In particular, our dataset caters more diverse and complex interactions in dense urban scenarios compared to the existing datasets. To address the challenges in predicting future trajectories with dense interactions, we develop a joint inference model that learns an expressive multi-modal shared latent space across agents in the urban scene. This enables our Joint-$\beta$-cVAE approach to better model the distribution of future trajectories. We achieve state of the art results on the nuScenes and Euro-PVI datasets demonstrating the importance of capturing interactions between ego-vehicle and pedestrians (bicyclists) for accurate predictions.

(2 min) Current popular phishing prevention techniques mainly utilize reactive blocklists, which leave a ``window of opportunity'' for attackers during which victims are unprotected. One possible approach to shorten this window aims to detect phishing attacks earlier, during website preparation, by monitoring Certificate Transparency (CT) logs. Previous attempts to work with CT log data for phishing classification exist, however they lack evaluations on actual CT log data. In this paper, we present a pipeline that facilitates such evaluations by addressing a number of problems when working with CT log data. The pipeline includes dataset creation, training, and past or live classification of CT logs. Its modular structure makes it possible to easily exchange classifiers or verification sources to support ground truth labeling efforts and classifier comparisons. We test the pipeline on a number of new and existing classifiers, and find a general potential to improve classifiers for this scenario in the future. We publish the source code of the pipeline and the used datasets along with this paper (https://gitlab.com/rwth-itsec/ctl-pipeline), thus making future research in this direction more accessible.

(2 min) We present a real-time neural radiance caching method for path-traced global illumination. Our system is designed to handle fully dynamic scenes, and makes no assumptions about the lighting, geometry, and materials. The data-driven nature of our approach sidesteps many difficulties of caching algorithms, such as locating, interpolating, and updating cache points. Since pretraining neural networks to handle novel, dynamic scenes is a formidable generalization challenge, we do away with pretraining and instead achieve generalization via adaptation, i.e. we opt for training the radiance cache while rendering. We employ self-training to provide low-noise training targets and simulate infinite-bounce transport by merely iterating few-bounce training updates. The updates and cache queries incur a mild overhead -- about 2.6ms on full HD resolution -- thanks to a streaming implementation of the neural network that fully exploits modern hardware. We demonstrate significant noise reduction at the cost of little induced bias, and report state-of-the-art, real-time performance on a number of challenging scenarios.

(2 min) We observe that despite their hierarchical convolutional nature, the synthesis process of typical generative adversarial networks depends on absolute pixel coordinates in an unhealthy manner. This manifests itself as, e.g., detail appearing to be glued to image coordinates instead of the surfaces of depicted objects. We trace the root cause to careless signal processing that causes aliasing in the generator network. Interpreting all signals in the network as continuous, we derive generally applicable, small architectural changes that guarantee that unwanted information cannot leak into the hierarchical synthesis process. The resulting networks match the FID of StyleGAN2 but differ dramatically in their internal representations, and they are fully equivariant to translation and rotation even at subpixel scales. Our results pave the way for generative models better suited for video and animation.

(2 min) We prove concentration inequalities for functions of independent random variables {under} sub-gaussian and sub-exponential conditions. The utility of the inequalities is demonstrated by an extension of the now classical method of Rademacher complexities to Lipschitz function classes and unbounded sub-exponential distribution.

(2 min) One widely used approach towards understanding the inner workings of deep convolutional neural networks is to visualize unit responses via activation maximization. Feature visualizations via activation maximization are thought to provide humans with precise information about the image features that cause a unit to be activated. If this is indeed true, these synthetic images should enable humans to predict the effect of an intervention, such as whether occluding a certain patch of the image (say, a dog's head) changes a unit's activation. Here, we test this hypothesis by asking humans to predict which of two square occlusions causes a larger change to a unit's activation. Both a large-scale crowdsourced experiment and measurements with experts show that on average, the extremely activating feature visualizations by Olah et al. (2017) indeed help humans on this task ($67 \pm 4\%$ accuracy; baseline performance without any visualizations is $60 \pm 3\%$). However, they do not provide any significant advantage over other visualizations (such as e.g. dataset samples), which yield similar performance ($66 \pm 3\%$ to $67 \pm 3\%$ accuracy). Taken together, we propose an objective psychophysical task to quantify the benefit of unit-level interpretability methods for humans, and find no evidence that feature visualizations provide humans with better "causal understanding" than simple alternative visualizations.

(2 min) We show that learning can be improved by using loss functions that evolve cyclically during training to emphasize one class at a time. In underparameterized networks, such dynamical loss functions can lead to successful training for networks that fail to find a deep minima of the standard cross-entropy loss. In overparameterized networks, dynamical loss functions can lead to better generalization. Improvement arises from the interplay of the changing loss landscape with the dynamics of the system as it evolves to minimize the loss. In particular, as the loss function oscillates, instabilities develop in the form of bifurcation cascades, which we study using the Hessian and Neural Tangent Kernel. Valleys in the landscape widen and deepen, and then narrow and rise as the loss landscape changes during a cycle. As the landscape narrows, the learning rate becomes too large and the network becomes unstable and bounces around the valley. This process ultimately pushes the system into deeper and wider regions of the loss landscape and is characterized by decreasing eigenvalues of the Hessian. This results in better regularized models with improved generalization performance.

(2 min) The idea of Innovation Search, which was initially proposed for data clustering, was recently used for outlier detection. In the application of Innovation Search for outlier detection, the directions of innovation were utilized to measure the innovation of the data points. We study the Innovation Values computed by the Innovation Search algorithm under a quadratic cost function and it is proved that Innovation Values with the new cost function are equivalent to Leverage Scores. This interesting connection is utilized to establish several theoretical guarantees for a Leverage Score based robust PCA method and to design a new robust PCA method. The theoretical results include performance guarantees with different models for the distribution of outliers and the distribution of inliers. In addition, we demonstrate the robustness of the algorithms against the presence of noise. The numerical and theoretical studies indicate that while the presented approach is fast and closed-form, it can outperform most of the existing algorithms.

(2 min) Numerous malware families rely on domain generation algorithms (DGAs) to establish a connection to their command and control (C2) server. Counteracting DGAs, several machine learning classifiers have been proposed enabling the identification of the DGA that generated a specific domain name and thus triggering targeted remediation measures. However, the proposed state-of-the-art classifiers are based on deep learning models. The black box nature of these makes it difficult to evaluate their reasoning. The resulting lack of confidence makes the utilization of such models impracticable. In this paper, we propose EXPLAIN, a feature-based and contextless DGA multiclass classifier. We comparatively evaluate several combinations of feature sets and hyperparameters for our approach against several state-of-the-art classifiers in a unified setting on the same real-world data. Our classifier achieves competitive results, is real-time capable, and its predictions are easier to trace back to features than the predictions made by the DGA multiclass classifiers proposed in related work.

(2 min) Multi-source unsupervised domain adaptation (MUDA) is a recently explored learning framework, where the goal is to address the challenge of labeled data scarcity in a target domain via transferring knowledge from multiple source domains with annotated data. Since the source data is distributed, the privacy of source domains' data can be a natural concern. We benefit from the idea of domain alignment in an embedding space to address the privacy concern for MUDA. Our method is based on aligning the sources and target distributions indirectly via internally learned distributions, without communicating data samples between domains. We justify our approach theoretically and perform extensive experiments to demonstrate that our method is effective and compares favorably against existing methods.

(2 min) We propose a new method for unsupervised continual knowledge consolidation in generative models that relies on the partitioning of Variational Autoencoder's latent space. Acquiring knowledge about new data samples without forgetting previous ones is a critical problem of continual learning. Currently proposed methods achieve this goal by extending the existing model while constraining its behavior not to degrade on the past data, which does not exploit the full potential of relations within the entire training dataset. In this work, we identify this limitation and posit the goal of continual learning as a knowledge accumulation task. We solve it by continuously re-aligning latent space partitions that we call bands which are representations of samples seen in different tasks, driven by the similarity of the information they contain. In addition, we introduce a simple yet effective method for controlled forgetting of past data that improves the quality of reconstructions encoded in latent bands and a latent space disentanglement technique that improves knowledge consolidation. On top of the standard continual learning evaluation benchmarks, we evaluate our method on a new knowledge consolidation scenario and show that the proposed approach outperforms state-of-the-art by up to twofold across all testing scenarios.

(2 min) Transformers recently are adapted from the community of natural language processing as a promising substitute of convolution-based neural networks for visual learning tasks. However, its supremacy degenerates given an insufficient amount of training data (e.g., ImageNet). To make it into practical utility, we propose a novel distillation-based method to train vision transformers. Unlike previous works, where merely heavy convolution-based teachers are provided, we introduce lightweight teachers with different architectural inductive biases (e.g., convolution and involution) to co-advise the student transformer. The key is that teachers with different inductive biases attain different knowledge despite that they are trained on the same dataset, and such different knowledge compounds and boosts the student's performance during distillation. Equipped with this cross inductive bias distillation method, our vision transformers (termed as CivT) outperform all previous transformers of the same architecture on ImageNet.

(2 min) Detecting the newly emerging malware variants in real time is crucial for mitigating cyber risks and proactively blocking intrusions. In this paper, we propose MG-DVD, a novel detection framework based on dynamic heterogeneous graph learning, to detect malware variants in real time. Particularly, MG-DVD first models the fine-grained execution event streams of malware variants into dynamic heterogeneous graphs and investigates real-world meta-graphs between malware objects, which can effectively characterize more discriminative malicious evolutionary patterns between malware and their variants. Then, MG-DVD presents two dynamic walk-based heterogeneous graph learning methods to learn more comprehensive representations of malware variants, which significantly reduces the cost of the entire graph retraining. As a result, MG-DVD is equipped with the ability to detect malware variants in real time, and it presents better interpretability by introducing meaningful meta-graphs. Comprehensive experiments on large-scale samples prove that our proposed MG-DVD outperforms state-of-the-art methods in detecting malware variants in terms of effectiveness and efficiency.

(2 min) This paper studies regret minimization in multi-armed bandits, a classical online learning problem. To develop more statistically-efficient algorithms, we propose to use the assumption of a random-effect model. In this model, the mean rewards of arms are drawn independently from an unknown distribution, whose parameters we estimate. We provide an estimator of the arm means in this model and also analyze its uncertainty. Based on these results, we design a UCB algorithm, which we call ReUCB. We analyze ReUCB and prove a Bayes regret bound on its $n$-round regret, which matches an existing lower bound. Our experiments show that ReUCB can outperform Thompson sampling in various scenarios, without assuming that the prior distribution of arm means is known.

(2 min) In this paper, we design a novel Bregman gradient policy optimization framework for reinforcement learning based on Bregman divergences and momentum techniques. Specifically, we propose a Bregman gradient policy optimization (BGPO) algorithm based on the basic momentum technique and mirror descent iteration. At the same time, we present an accelerated Bregman gradient policy optimization (VR-BGPO) algorithm based on a momentum variance-reduced technique. Moreover, we introduce a convergence analysis framework for our Bregman gradient policy optimization under the nonconvex setting. Specifically, we prove that BGPO achieves the sample complexity of $\tilde{O}(\epsilon^{-4})$ for finding $\epsilon$-stationary point only requiring one trajectory at each iteration, and VR-BGPO reaches the best known sample complexity of $\tilde{O}(\epsilon^{-3})$ for finding an $\epsilon$-stationary point, which also only requires one trajectory at each iteration. In particular, by using different Bregman divergences, our methods unify many existing policy optimization algorithms and their new variants such as the existing (variance-reduced) policy gradient algorithms and (variance-reduced) natural policy gradient algorithms. Extensive experimental results on multiple reinforcement learning tasks demonstrate the efficiency of our new algorithms.

(2 min) We provide a setting and a general approach to fair online learning with stochastic sensitive and non-sensitive contexts. The setting is a repeated game between the Player and Nature, where at each stage both pick actions based on the contexts. Inspired by the notion of unawareness, we assume that the Player can only access the non-sensitive context before making a decision, while we discuss both cases of Nature accessing the sensitive contexts and Nature unaware of the sensitive contexts. Adapting Blackwell's approachability theory to handle the case of an unknown contexts' distribution, we provide a general necessary and sufficient condition for learning objectives to be compatible with some fairness constraints. This condition is instantiated on (group-wise) no-regret and (group-wise) calibration objectives, and on demographic parity as an additional constraint. When the objective is not compatible with the constraint, the provided framework permits to characterise the optimal trade-off between the two.

(2 min) We consider estimation and control in linear time-varying dynamical systems from the perspective of regret minimization. Unlike most prior work in this area, we focus on the problem of designing causal estimators and controllers which compete against a clairvoyant noncausal policy, instead of the best policy selected in hindsight from some fixed parametric class. We show that the regret-optimal estimator and regret-optimal controller can be derived in state-space form using operator-theoretic techniques from robust control and present tight,data-dependent bounds on the regret incurred by our algorithms in terms of the energy of the disturbances. Our results can be viewed as extending traditional robust estimation and control, which focuses on minimizing worst-case cost, to minimizing worst-case regret. We propose regret-optimal analogs of Model-Predictive Control (MPC) and the Extended KalmanFilter (EKF) for systems with nonlinear dynamics and present numerical experiments which show that our regret-optimal algorithms can significantly outperform standard approaches to estimation and control.

(2 min) Bayesian networks are probabilistic graphical models with a wide range of application areas including gene regulatory networks inference, risk analysis and image processing. Learning the structure of a Bayesian network (BNSL) from discrete data is known to be an NP-hard task with a superexponential search space of directed acyclic graphs. In this work, we propose a new polynomial time algorithm for discovering a subset of all possible cluster cuts, a greedy algorithm for approximately solving the resulting linear program, and a generalised arc consistency algorithm for the acyclicity constraint. We embed these in the constraint programmingbased branch-and-bound solver CPBayes and show that, despite being suboptimal, they improve performance by orders of magnitude. The resulting solver also compares favourably with GOBNILP, a state-of-the-art solver for the BNSL problem which solves an NP-hard problem to discover each cut and solves the linear program exactly.

(2 min) Cloud computing service models have experienced rapid growth and inefficient resource usage is known as one of the greatest causes of high energy consumption in cloud data centers. Resource allocation in cloud data centers aiming to reduce energy consumption has been conducted using live migration of Virtual Machines (VMs) and their consolidation into the small number of Physical Machines (PMs). However, the selection of the appropriate VM for migration is an important challenge. To solve this issue, VMs can be classified according to the pattern of user requests into sensitive or insensitive classes to latency, and thereafter suitable VMs can be selected for migration. In this paper, the combination of Convolution Neural Network (CNN) and Gated Recurrent Unit (GRU) is utilized for the classification of VMs in the Microsoft Azure dataset. Due to the fact the majority of VMs in this dataset are labeled as insensitive to latency, migration of more VMs in this group not only reduces energy consumption but also decreases the violation of Service Level Agreements (SLA). Based on the empirical results, the proposed model obtained an accuracy of 95.18which clearly demonstrates the superiority of our proposed model compared to other existing models.

(2 min) Supercomputer FCFS-based scheduling policies result in many transient idle nodes, a phenomenon that is only partially alleviated by backfill scheduling methods that promote small jobs to run before large jobs. Here we describe how to realize a novel use for these otherwise wasted resources, namely, deep neural network (DNN) training. This important workload is easily organized as many small fragments that can be configured dynamically to fit essentially any node*time hole in a supercomputer's schedule. We describe how the task of rescaling suitable DNN training tasks to fit dynamically changing holes can be formulated as a deterministic mixed integer linear programming (MILP)-based resource allocation algorithm, and show that this MILP problem can be solved efficiently at run time. We show further how this MILP problem can be adapted to optimize for administrator- or user-defined metrics. We validate our method with supercomputer scheduler logs and different DNN training scenarios, and demonstrate efficiencies of up to 93% compared with running the same training tasks on dedicated nodes. Our method thus enables substantial supercomputer resources to be allocated to DNN training with no impact on other applications.

(2 min) Transfer learning is essential when sufficient data comes from the source domain, with scarce labeled data from the target domain. We develop estimators that achieve minimax linear risk for linear regression problems under distribution shift. Our algorithms cover different transfer learning settings including covariate shift and model shift. We also consider when data are generated from either linear or general nonlinear models. We show that linear minimax estimators are within an absolute constant of the minimax risk even among nonlinear estimators for various source/target distributions.

(2 min) This work is devoted to the clustering of check-in sequences from a geosocial network. We used the mixture Markov chain process as a mathematical model for time-dependent types of data. For clustering, we adjusted the Expectation-Maximization (EM) algorithm. As a result, we obtained highly detailed communities (clusters) of users of the now defunct geosocial network, Weeplaces.

(2 min) Deep neural networks are behind many of the recent successes in machine learning applications. However, these models can produce overconfident decisions while encountering out-of-distribution (OOD) examples or making a wrong prediction. This inconsistent predictive confidence limits the integration of independently-trained learning models into a larger system. This paper introduces separable concept learning framework to realistically measure the performance of classifiers in presence of OOD examples. In this setup, several instances of a classifier are trained on different parts of a partition of the set of classes. Later, the performance of the combination of these models is evaluated on a separate test set. Unlike current OOD detection techniques, this framework does not require auxiliary OOD datasets and does not separate classification from detection performance. Furthermore, we present a new strong baseline for more consistent predictive confidence in deep models, called fitted ensembles, where overconfident predictions are rectified by transformed versions of the original classification task. Fitted ensembles can naturally detect OOD examples without requiring auxiliary data by observing contradicting predictions among its components. Experiments on MNIST, SVHN, CIFAR-10/100, and ImageNet show fitted ensemble significantly outperform conventional ensembles on OOD examples and are possible to scale.

(2 min) We introduce ParK, a new large-scale solver for kernel ridge regression. Our approach combines partitioning with random projections and iterative optimization to reduce space and time complexity while provably maintaining the same statistical accuracy. In particular, constructing suitable partitions directly in the feature space rather than in the input space, we promote orthogonality between the local estimators, thus ensuring that key quantities such as local effective dimension and bias remain under control. We characterize the statistical-computational tradeoff of our model, and demonstrate the effectiveness of our method by numerical experiments on large-scale datasets.

(2 min) Conventional representation learning algorithms for knowledge graphs (KG) map each entity to a unique embedding vector. Such a shallow lookup results in a linear growth of memory consumption for storing the embedding matrix and incurs high computational costs when working with real-world KGs. Drawing parallels with subword tokenization commonly used in NLP, we explore the landscape of more parameter-efficient node embedding strategies with possibly sublinear memory requirements. To this end, we propose NodePiece, an anchor-based approach to learn a fixed-size entity vocabulary. In NodePiece, a vocabulary of subword/sub-entity units is constructed from anchor nodes in a graph with known relation types. Given such a fixed-size vocabulary, it is possible to bootstrap an encoding and embedding for any entity, including those unseen during training. Experiments show that NodePiece performs competitively in node classification, link prediction, and relation prediction tasks while retaining less than 10% of explicit nodes in a graph as anchors and often having 10x fewer parameters.

(2 min) This work tackles the issue of fairness in the context of generative procedures, such as image super-resolution, which entail different definitions from the standard classification setting. Moreover, while traditional group fairness definitions are typically defined with respect to specified protected groups -- camouflaging the fact that these groupings are artificial and carry historical and political motivations -- we emphasize that there are no ground truth identities. For instance, should South and East Asians be viewed as a single group or separate groups? Should we consider one race as a whole or further split by gender? Choosing which groups are valid and who belongs in them is an impossible dilemma and being ``fair'' with respect to Asians may require being ``unfair'' with respect to South Asians. This motivates the introduction of definitions that allow algorithms to be \emph{oblivious} to the relevant groupings. We define several intuitive notions of group fairness and study their incompatibilities and trade-offs. We show that the natural extension of demographic parity is strongly dependent on the grouping, and \emph{impossible} to achieve obliviously. On the other hand, the conceptually new definition we introduce, Conditional Proportional Representation, can be achieved obliviously through Posterior Sampling. Our experiments validate our theoretical results and achieve fair image reconstruction using state-of-the-art generative models.

(2 min) Atomic clauses are fundamental text units for understanding complex sentences. Identifying the atomic sentences within complex sentences is important for applications such as summarization, argument mining, discourse analysis, discourse parsing, and question answering. Previous work mainly relies on rule-based methods dependent on parsing. We propose a new task to decompose each complex sentence into simple sentences derived from the tensed clauses in the source, and a novel problem formulation as a graph edit task. Our neural model learns to Accept, Break, Copy or Drop elements of a graph that combines word adjacency and grammatical dependencies. The full processing pipeline includes modules for graph construction, graph editing, and sentence generation from the output graph. We introduce DeSSE, a new dataset designed to train and evaluate complex sentence decomposition, and MinWiki, a subset of MinWikiSplit. ABCD achieves comparable performance as two parsing baselines on MinWiki. On DeSSE, which has a more even balance of complex sentence types, our model achieves higher accuracy on the number of atomic sentences than an encoder-decoder baseline. Results include a detailed error analysis.

(2 min) Gaussian processes are one of the dominant approaches in Bayesian learning. Although the approach has been applied to numerous problems with great success, it has a few fundamental limitations. Multiple methods in literature have addressed these limitations. However, there has not been a comprehensive survey of the topics as of yet. Most existing surveys focus on only one particular variant of Gaussian processes and their derivatives. This survey details the core motivations for using Gaussian processes, their mathematical formulations, limitations, and research themes that have flourished over the years to address said limitations. Furthermore, one particular research area is Deep Gaussian Processes (DGPs), it has improved substantially in the past decade. The significant publications that advanced the forefront of this research area are outlined in their survey. Finally, a brief discussion on open problems and research directions for future work is presented at the end.

(2 min) To further improve the learning efficiency and performance of reinforcement learning (RL), in this paper we propose a novel uncertainty-aware model-based RL (UA-MBRL) framework, and then implement and validate it in autonomous driving under various task scenarios. First, an action-conditioned ensemble model with the ability of uncertainty assessment is established as the virtual environment model. Then, a novel uncertainty-aware model-based RL framework is developed based on the adaptive truncation approach, providing virtual interactions between the agent and environment model, and improving RL's training efficiency and performance. The developed algorithms are then implemented in end-to-end autonomous vehicle control tasks, validated and compared with state-of-the-art methods under various driving scenarios. The validation results suggest that the proposed UA-MBRL method surpasses the existing model-based and model-free RL approaches, in terms of learning efficiency and achieved performance. The results also demonstrate the good ability of the proposed method with respect to the adaptiveness and robustness, under various autonomous driving scenarios.

(2 min) Multi-action restless multi-armed bandits (RMABs) are a powerful framework for constrained resource allocation in which $N$ independent processes are managed. However, previous work only study the offline setting where problem dynamics are known. We address this restrictive assumption, designing the first algorithms for learning good policies for Multi-action RMABs online using combinations of Lagrangian relaxation and Q-learning. Our first approach, MAIQL, extends a method for Q-learning the Whittle index in binary-action RMABs to the multi-action setting. We derive a generalized update rule and convergence proof and establish that, under standard assumptions, MAIQL converges to the asymptotically optimal multi-action RMAB policy as $t\rightarrow{}\infty$. However, MAIQL relies on learning Q-functions and indexes on two timescales which leads to slow convergence and requires problem structure to perform well. Thus, we design a second algorithm, LPQL, which learns the well-performing and more general Lagrange policy for multi-action RMABs by learning to minimize the Lagrange bound through a variant of Q-learning. To ensure fast convergence, we take an approximation strategy that enables learning on a single timescale, then give a guarantee relating the approximation's precision to an upper bound of LPQL's return as $t\rightarrow{}\infty$. Finally, we show that our approaches always outperform baselines across multiple settings, including one derived from real-world medication adherence data.

(2 min) Many machine learning techniques incorporate identity-preserving transformations into their models to generalize their performance to previously unseen data. These transformations are typically selected from a set of functions that are known to maintain the identity of an input when applied (e.g., rotation, translation, flipping, and scaling). However, there are many natural variations that cannot be labeled for supervision or defined through examination of the data. As suggested by the manifold hypothesis, many of these natural variations live on or near a low-dimensional, nonlinear manifold. Several techniques represent manifold variations through a set of learned Lie group operators that define directions of motion on the manifold. However theses approaches are limited because they require transformation labels when training their models and they lack a method for determining which regions of the manifold are appropriate for applying each specific operator. We address these limitations by introducing a learning strategy that does not require transformation labels and developing a method that learns the local regions where each operator is likely to be used while preserving the identity of inputs. Experiments on MNIST and Fashion MNIST highlight our model's ability to learn identity-preserving transformations on multi-class datasets. Additionally, we train on CelebA to showcase our model's ability to learn semantically meaningful transformations on complex datasets in an unsupervised manner.

(2 min) Information theory is of importance to machine learning, but the notation for information-theoretic quantities is sometimes opaque. The right notation can convey valuable intuitions and concisely express new ideas. We propose such a notation for machine learning users and expand it to include information-theoretic quantities between events (outcomes) and random variables. We apply this notation to a popular information-theoretic acquisition function in Bayesian active learning which selects the most informative (unlabelled) samples to be labelled by an expert. We demonstrate the value of our notation when extending the acquisition function to the core-set problem, which consists of selecting the most informative samples \emph{given} the labels.

(2 min) Commonsense reasoning is one of the key problems in natural language processing, but the relative scarcity of labeled data holds back the progress for languages other than English. Pretrained cross-lingual models are a source of powerful language-agnostic representations, yet their inherent reasoning capabilities are still actively studied. In this work, we design a simple approach to commonsense reasoning which trains a linear classifier with weights of multi-head attention as features. To evaluate this approach, we create a multilingual Winograd Schema corpus by processing several datasets from prior work within a standardized pipeline and measure cross-lingual generalization ability in terms of out-of-sample performance. The method performs competitively with recent supervised and unsupervised approaches for commonsense reasoning, even when applied to other languages in a zero-shot manner. Also, we demonstrate that most of the performance is given by the same small subset of attention heads for all studied languages, which provides evidence of universal reasoning capabilities in multilingual encoders.

(2 min) The formation of precipitation in state-of-the-art weather and climate models is an important process. The understanding of its relationship with other variables can lead to endless benefits, particularly for the world's monsoon regions dependent on rainfall as a support for livelihood. Various factors play a crucial role in the formation of rainfall, and those physical processes are leading to significant biases in the operational weather forecasts. We use the UNET architecture of a deep convolutional neural network with residual learning as a proof of concept to learn global data-driven models of precipitation. The models are trained on reanalysis datasets projected on the cubed-sphere projection to minimize errors due to spherical distortion. The results are compared with the operational dynamical model used by the India Meteorological Department. The theoretical deep learning-based model shows doubling of the grid point, as well as area averaged skill measured in Pearson correlation coefficients relative to operational system. This study is a proof-of-concept showing that residual learning-based UNET can unravel physical relationships to target precipitation, and those physical constraints can be used in the dynamical operational models towards improved precipitation forecasts. Our results pave the way for the development of online, hybrid models in the future.

(2 min) We study pure exploration in bandits, where the dimension of the feature representation can be much larger than the number of arms. To overcome the curse of dimensionality, we propose to adaptively embed the feature representation of each arm into a lower-dimensional space and carefully deal with the induced model misspecifications. Our approach is conceptually very different from existing works that can either only handle low-dimensional linear bandits or passively deal with model misspecifications. We showcase the application of our approach to two pure exploration settings that were previously under-studied: (1) the reward function belongs to a possibly infinite-dimensional Reproducing Kernel Hilbert Space, and (2) the reward function is nonlinear and can be approximated by neural networks. Our main results provide sample complexity guarantees that only depend on the effective dimension of the feature spaces in the kernel or neural representations. Extensive experiments conducted on both synthetic and real-world datasets demonstrate the efficacy of our methods.

(2 min) Climate models are critical tools for developing strategies to manage the risks posed by sea-level rise to coastal communities. While these models are necessary for understanding climate risks, there is a level of uncertainty inherent in each parameter in the models. This model parametric uncertainty leads to uncertainty in future climate risks. Consequently, there is a need to understand how those parameter uncertainties impact our assessment of future climate risks and the efficacy of strategies to manage them. Here, we use random forests to examine the parametric drivers of future climate risk and how the relative importances of those drivers change over time. We find that the equilibrium climate sensitivity and a factor that scales the effect of aerosols on radiative forcing are consistently the most important climate model parametric uncertainties throughout the 2020 to 2150 interval for both low and high radiative forcing scenarios. The near-term hazards of high-end sea-level rise are driven primarily by thermal expansion, while the longer-term hazards are associated with mass loss from the Antarctic and Greenland ice sheets. Our results highlight the practical importance of considering time-evolving parametric uncertainties when developing strategies to manage future climate risks.

(2 min) In active learning, new labels are commonly acquired in batches. However, common acquisition functions are only meant for one-sample acquisition rounds at a time, and when their scores are used naively for batch acquisition, they result in batches lacking diversity, which deteriorates performance. On the other hand, state-of-the-art batch acquisition functions are costly to compute. In this paper, we present a novel class of stochastic acquisition functions that extend one-sample acquisition functions to the batch setting by observing how one-sample acquisition scores change as additional samples are acquired and modelling this difference for additional batch samples. We simply acquire new samples by sampling from the pool set using a Gibbs distribution based on the acquisition scores. Our acquisition functions are both vastly cheaper to compute and out-perform other batch acquisition functions.

(2 min) Many learning-based 3D shape semantic segmentation methods assign labels to shape atoms (e.g. points in a point cloud or faces in a mesh) with a single-pass approach trained in an end-to-end fashion. Such methods achieve impressive performance but require large amounts of labeled training data. This paradigm entangles two separable subproblems: (1) decomposing a shape into regions and (2) assigning semantic labels to these regions. We claim that disentangling these subproblems reduces the labeled data burden: (1) region decomposition requires no semantic labels and could be performed in an unsupervised fashion, and (2) labeling shape regions instead of atoms results in a smaller search space and should be learnable with less labeled training data. In this paper, we investigate this second claim by presenting the Neurally-Guided Shape Parser (NGSP), a method that learns how to assign semantic labels to regions of an over-segmented 3D shape. We solve this problem via MAP inference, modeling the posterior probability of a labeling assignment conditioned on an input shape. We employ a Monte Carlo importance sampling approach guided by a neural proposal network, a search-based approach made feasible by assuming the input shape is decomposed into discrete regions. We evaluate NGSP on the task of hierarchical semantic segmentation on manufactured 3D shapes from PartNet. We find that NGSP delivers significant performance improvements over baselines that learn to label shape atoms and then aggregate predictions for each shape region, especially in low-data regimes. Finally, we demonstrate that NGSP is robust to region granularity, as it maintains strong segmentation performance even as the regions undergo significant corruption.

(2 min) Due to high annotation costs making the best use of existing human-created training data is an important research direction. We, therefore, carry out a systematic evaluation of transferability of BERT-based neural ranking models across five English datasets. Previous studies focused primarily on zero-shot and few-shot transfer from a large dataset to a dataset with a small number of queries. In contrast, each of our collections has a substantial number of queries, which enables a full-shot evaluation mode and improves reliability of our results. Furthermore, since source datasets licences often prohibit commercial use, we compare transfer learning to training on pseudo-labels generated by a BM25 scorer. We find that training on pseudo-labels -- possibly with subsequent fine-tuning using a modest number of annotated queries -- can produce a competitive or better model compared to transfer learning. Yet, it is necessary to improve the stability and/or effectiveness of the few-shot training, which, sometimes, can degrade performance of a pretrained model.

(2 min) Distant supervision for relation extraction provides uniform bag labels for each sentence inside the bag, while accurate sentence labels are important for downstream applications that need the exact relation type. Directly using bag labels for sentence-level training will introduce much noise, thus severely degrading performance. In this work, we propose the use of negative training (NT), in which a model is trained using complementary labels regarding that ``the instance does not belong to these complementary labels". Since the probability of selecting a true label as a complementary label is low, NT provides less noisy information. Furthermore, the model trained with NT is able to separate the noisy data from the training data. Based on NT, we propose a sentence-level framework, SENT, for distant relation extraction. SENT not only filters the noisy data to construct a cleaner dataset, but also performs a re-labeling process to transform the noisy data into useful training data, thus further benefiting the model's performance. Experimental results show the significant improvement of the proposed method over previous methods on sentence-level evaluation and de-noise effect.

(2 min) Plug-and-play language models (PPLMs) enable topic-conditioned natural language generation by pairing large pre-trained generators with attribute models used to steer the predicted token distribution towards the selected topic. Despite their computational efficiency, PPLMs require large amounts of labeled texts to effectively balance generation fluency and proper conditioning, making them unsuitable for low-resource settings. We present ETC-NLG, an approach leveraging topic modeling annotations to enable fully-unsupervised End-to-end Topic-Conditioned Natural Language Generation over emergent topics in unlabeled document collections. We first test the effectiveness of our approach in a low-resource setting for Italian, evaluating the conditioning for both topic models and gold annotations. We then perform a comparative evaluation of ETC-NLG for Italian and English using a parallel corpus. Finally, we propose an automatic approach to estimate the effectiveness of conditioning on the generated utterances.

(2 min) Natural Language Processing (NLP) and Information Retrieval (IR) in the judicial domain is an essential task. With the advent of availability domain-specific data in electronic form and aid of different Artificial intelligence (AI) technologies, automated language processing becomes more comfortable, and hence it becomes feasible for researchers and developers to provide various automated tools to the legal community to reduce human burden. The Competition on Legal Information Extraction/Entailment (COLIEE-2019) run in association with the International Conference on Artificial Intelligence and Law (ICAIL)-2019 has come up with few challenging tasks. The shared defined four sub-tasks (i.e. Task1, Task2, Task3 and Task4), which will be able to provide few automated systems to the judicial system. The paper presents our working note on the experiments carried out as a part of our participation in all the sub-tasks defined in this shared task. We make use of different Information Retrieval(IR) and deep learning based approaches to tackle these problems. We obtain encouraging results in all these four sub-tasks.

(2 min) Now, the pre-training technique is ubiquitous in natural language processing field. ProphetNet is a pre-training based natural language generation method which shows powerful performance on English text summarization and question generation tasks. In this paper, we extend ProphetNet into other domains and languages, and present the ProphetNet family pre-training models, named ProphetNet-X, where X can be English, Chinese, Multi-lingual, and so on. We pre-train a cross-lingual generation model ProphetNet-Multi, a Chinese generation model ProphetNet-Zh, two open-domain dialog generation models ProphetNet-Dialog-En and ProphetNet-Dialog-Zh. And also, we provide a PLG (Programming Language Generation) model ProphetNet-Code to show the generation performance besides NLG (Natural Language Generation) tasks. In our experiments, ProphetNet-X models achieve new state-of-the-art performance on 10 benchmarks. All the models of ProphetNet-X share the same model structure, which allows users to easily switch between different models. We make the code and models publicly available, and we will keep updating more pre-training models and finetuning scripts.

(2 min) Modern pre-trained language models are mostly built upon backbones stacking self-attention and feed-forward layers in an interleaved order. In this paper, beyond this stereotyped layer pattern, we aim to improve pre-trained models by exploiting layer variety from two aspects: the layer type set and the layer order. Specifically, besides the original self-attention and feed-forward layers, we introduce convolution into the layer type set, which is experimentally found beneficial to pre-trained models. Furthermore, beyond the original interleaved order, we explore more layer orders to discover more powerful architectures. However, the introduced layer variety leads to a large architecture space of more than billions of candidates, while training a single candidate model from scratch already requires huge computation cost, making it not affordable to search such a space by directly training large amounts of candidate models. To solve this problem, we first pre-train a supernet from which the weights of all candidate models can be inherited, and then adopt an evolutionary algorithm guided by pre-training accuracy to find the optimal architecture. Extensive experiments show that LV-BERT model obtained by our method outperforms BERT and its variants on various downstream tasks. For example, LV-BERT-small achieves 78.8 on the GLUE testing set, 1.8 higher than the strong baseline ELECTRA-small.

(2 min) Tackling online hatred using informed textual responses - called counter narratives - has been brought under the spotlight recently. Accordingly, a research line has emerged to automatically generate counter narratives in order to facilitate the direct intervention in the hate discussion and to prevent hate content from further spreading. Still, current neural approaches tend to produce generic/repetitive responses and lack grounded and up-to-date evidence such as facts, statistics, or examples. Moreover, these models can create plausible but not necessarily true arguments. In this paper we present the first complete knowledge-bound counter narrative generation pipeline, grounded in an external knowledge repository that can provide more informative content to fight online hatred. Together with our approach, we present a series of experiments that show its feasibility to produce suitable and informative counter narratives in in-domain and cross-domain settings.

(2 min) In this work, we train the first monolingual Lithuanian transformer model on a relatively large corpus of Lithuanian news articles and compare various output decoding algorithms for abstractive news summarization. We achieve an average ROUGE-2 score 0.163, generated summaries are coherent and look impressive at first glance. However, some of them contain misleading information that is not so easy to spot. We describe all the technical details and share our trained model and accompanying code in an online open-source repository, as well as some characteristic samples of the generated summaries.

(2 min) The majority of existing methods for empathetic response generation rely on the emotion of the context to generate empathetic responses. However, empathy is much more than generating responses with an appropriate emotion. It also often entails subtle expressions of understanding and personal resonance with the situation of the other interlocutor. Unfortunately, such qualities are difficult to quantify and the datasets lack the relevant annotations. To address this issue, in this paper we propose an approach that relies on exemplars to cue the generative model on fine stylistic properties that signal empathy to the interlocutor. To this end, we employ dense passage retrieval to extract relevant exemplary responses from the training set. Three elements of human communication -- emotional presence, interpretation, and exploration, and sentiment are additionally introduced using synthetic labels to guide the generation towards empathy. The human evaluation is also extended by these elements of human communication. We empirically show that these approaches yield significant improvements in empathetic response quality in terms of both automated and human-evaluated metrics. The implementation is available at https://github.com/declare-lab/exemplary-empathy.

(2 min) Cross-lingual adaptation with multilingual pre-trained language models (mPTLMs) mainly consists of two lines of works: zero-shot approach and translation-based approach, which have been studied extensively on the sequence-level tasks. We further verify the efficacy of these cross-lingual adaptation approaches by evaluating their performances on more fine-grained sequence tagging tasks. After re-examining their strengths and drawbacks, we propose a novel framework to consolidate the zero-shot approach and the translation-based approach for better adaptation performance. Instead of simply augmenting the source data with the machine-translated data, we tailor-make a warm-up mechanism to quickly update the mPTLMs with the gradients estimated on a few translated data. Then, the adaptation approach is applied to the refined parameters and the cross-lingual transfer is performed in a warm-start way. The experimental results on nine target languages demonstrate that our method is beneficial to the cross-lingual adaptation of various sequence tagging tasks.

(2 min) In semantic parsing of geographical queries against real-world databases such as OpenStreetMap (OSM), unique correct answers do not necessarily exist. Instead, the truth might be lying in the eye of the user, who needs to enter an interactive setup where ambiguities can be resolved and parsing mistakes can be corrected. Our work presents an approach to interactive semantic parsing where an explicit error detection is performed, and a clarification question is generated that pinpoints the suspected source of ambiguity or error and communicates it to the human user. Our experimental results show that a combination of entropy-based uncertainty detection and beam search, together with multi-source training on clarification question, initial parse, and user answer, results in improvements of 1.2% F1 score on a parser that already performs at 90.26% on the NLMaps dataset for OSM semantic parsing.

(2 min) Language models trained on billions of tokens have recently led to unprecedented results on many NLP tasks. This success raises the question of whether, in principle, a system can ever ``understand'' raw text without access to some form of grounding. We formally investigate the abilities of ungrounded systems to acquire meaning. Our analysis focuses on the role of ``assertions'': textual contexts that provide indirect clues about the underlying semantics. We study whether assertions enable a system to emulate representations preserving semantic relations like equivalence. We find that assertions enable semantic emulation of languages that satisfy a strong notion of semantic transparency. However, for classes of languages where the same expression can take different values in different contexts, we show that emulation can become uncomputable. Finally, we discuss differences between our formal model and natural language, exploring how our results generalize to a modal setting and other semantic relations. Together, our results suggest that assertions in code or language do not provide sufficient signal to fully emulate semantic representations. We formalize ways in which ungrounded language models appear to be fundamentally limited in their ability to ``understand''.

(2 min) Inspired by evidence that pretrained language models (LMs) encode commonsense knowledge, recent work has applied LMs to automatically populate commonsense knowledge graphs (CKGs). However, there is a lack of understanding on their generalization to multiple CKGs, unseen relations, and novel entities. This paper analyzes the ability of LMs to perform generalizable commonsense inference, in terms of knowledge capacity, transferability, and induction. Our experiments with these three aspects show that: (1) LMs can adapt to different schemas defined by multiple CKGs but fail to reuse the knowledge to generalize to new relations. (2) Adapted LMs generalize well to unseen subjects, but less so on novel objects. Future work should investigate how to improve the transferability and induction of commonsense mining from LMs.

(2 min) Most of the existing works for dialogue generation are data-driven models trained directly on corpora crawled from websites. They mainly focus on improving the model architecture to produce better responses but pay little attention to considering the quality of the training data contrastively. In this paper, we propose a multi-level contrastive learning paradigm to model the fine-grained quality of the responses with respect to the query. A Rank-aware Calibration (RC) network is designed to construct the multi-level contrastive optimization objectives. Since these objectives are calculated based on the sentence level, which may erroneously encourage/suppress the generation of uninformative/informative words. To tackle this incidental issue, on one hand, we design an exquisite token-level strategy for estimating the instance loss more accurately. On the other hand, we build a Knowledge Inference (KI) component to capture the keyword knowledge from the reference during training and exploit such information to encourage the generation of informative words. We evaluate the proposed model on a carefully annotated dialogue dataset and the results suggest that our model can generate more relevant and diverse responses compared to the baseline models.

(2 min) Routing methods in capsule networks often learn a hierarchical relationship for capsules in successive layers, but the intra-relation between capsules in the same layer is less studied, while this intra-relation is a key factor for the semantic understanding in text data. Therefore, in this paper, we introduce a new capsule network with graph routing to learn both relationships, where capsules in each layer are treated as the nodes of a graph. We investigate strategies to yield adjacency and degree matrix with three different distances from a layer of capsules, and propose the graph routing mechanism between those capsules. We validate our approach on five text classification datasets, and our findings suggest that the approach combining bottom-up routing and top-down attention performs the best. Such an approach demonstrates generalization capability across datasets. Compared to the state-of-the-art routing methods, the improvements in accuracy in the five datasets we used were 0.82, 0.39, 0.07, 1.01, and 0.02, respectively.

(2 min) One of the main challenges in conversational question answering (CQA) is to resolve the conversational dependency, such as anaphora and ellipsis. However, existing approaches do not explicitly train QA models on how to resolve the dependency, and thus these models are limited in understanding human dialogues. In this paper, we propose a novel framework, ExCorD (Explicit guidance on how to resolve Conversational Dependency) to enhance the abilities of QA models in comprehending conversational context. ExCorD first generates self-contained questions that can be understood without the conversation history, then trains a QA model with the pairs of original and self-contained questions using a consistency-based regularizer. In our experiments, we demonstrate that ExCorD significantly improves the QA models' performance by up to 1.2 F1 on QuAC, and 5.2 F1 on CANARD, while addressing the limitations of the existing approaches.

(2 min) Current task-oriented dialog (TOD) systems mostly manage structured knowledge (e.g. databases and tables) to guide the goal-oriented conversations. However, they fall short of handling dialogs which also involve unstructured knowledge (e.g. reviews and documents). In this paper, we formulate a task of modeling TOD grounded on a fusion of structured and unstructured knowledge. To address this task, we propose a TOD system with semi-structured knowledge management, SeKnow, which extends the belief state to manage knowledge with both structured and unstructured contents. Furthermore, we introduce two implementations of SeKnow based on a non-pretrained sequence-to-sequence model and a pretrained language model, respectively. Both implementations use the end-to-end manner to jointly optimize dialog modeling grounded on structured and unstructured knowledge. We conduct experiments on the modified version of MultiWOZ 2.1 dataset, where dialogs are processed to involve semi-structured knowledge. Experimental results show that SeKnow has strong performances in both end-to-end dialog and intermediate knowledge management, compared to existing TOD systems and their extensions with pipeline knowledge management schemes.

(2 min) Dysfluencies and variations in speech pronunciation can severely degrade speech recognition performance, and for many individuals with moderate-to-severe speech disorders, voice operated systems do not work. Current speech recognition systems are trained primarily with data from fluent speakers and as a consequence do not generalize well to speech with dysfluencies such as sound or word repetitions, sound prolongations, or audible blocks. The focus of this work is on quantitative analysis of a consumer speech recognition system on individuals who stutter and production-oriented approaches for improving performance for common voice assistant tasks (i.e., "what is the weather?"). At baseline, this system introduces a significant number of insertion and substitution errors resulting in intended speech Word Error Rates (isWER) that are 13.64\% worse (absolute) for individuals with fluency disorders. We show that by simply tuning the decoding parameters in an existing hybrid speech recognition system one can improve isWER by 24\% (relative) for individuals with fluency disorders. Tuning these parameters translates to 3.6\% better domain recognition and 1.7\% better intent recognition relative to the default setup for the 18 study participants across all stuttering severities.

(2 min) As neural machine translation (NMT) systems become an important part of professional translator pipelines, a growing body of work focuses on combining NMT with terminologies. In many scenarios and particularly in cases of domain adaptation, one expects the MT output to adhere to the constraints provided by a terminology. In this work, we propose metrics to measure the consistency of MT output with regards to a domain terminology. We perform studies on the COVID-19 domain over 5 languages, also performing terminology-targeted human evaluation. We open-source the code for computing all proposed metrics: https://github.com/mahfuzibnalam/terminology_evaluation

(2 min) A wide variety of NLP applications, such as machine translation, summarization, and dialog, involve text generation. One major challenge for these applications is how to evaluate whether such generated texts are actually fluent, accurate, or effective. In this work, we conceptualize the evaluation of generated text as a text generation problem, modeled using pre-trained sequence-to-sequence models. The general idea is that models trained to convert the generated text to/from a reference output or the source text will achieve higher scores when the generated text is better. We operationalize this idea using BART, an encoder-decoder based pre-trained model, and propose a metric BARTScore with a number of variants that can be flexibly applied in an unsupervised fashion to evaluation of text from different perspectives (e.g. informativeness, fluency, or factuality). BARTScore is conceptually simple and empirically effective. It can outperform existing top-scoring metrics in 16 of 22 test settings, covering evaluation of 16 datasets (e.g., machine translation, text summarization) and 7 different perspectives (e.g., informativeness, factuality). Code to calculate BARTScore is available at https://github.com/neulab/BARTScore, and we have released an interactive leaderboard for meta-evaluation at this http URL on the ExplainaBoard platform, which allows us to interactively understand the strengths, weaknesses, and complementarity of each metric.

(2 min) State-of-the-art summarization systems are trained and evaluated on massive datasets scraped from the web. Despite their prevalence, we know very little about the underlying characteristics (data noise, summarization complexity, etc.) of these datasets, and how these affect system performance and the reliability of automatic metrics like ROUGE. In this study, we manually analyze 600 samples from three popular summarization datasets. Our study is driven by a six-class typology which captures different noise types (missing facts, entities) and degrees of summarization difficulty (extractive, abstractive). We follow with a thorough analysis of 27 state-of-the-art summarization models and 5 popular metrics, and report our key insights: (1) Datasets have distinct data quality and complexity distributions, which can be traced back to their collection process. (2) The performance of models and reliability of metrics is dependent on sample complexity. (3) Faithful summaries often receive low scores because of the poor diversity of references. We release the code, annotated data and model outputs.

(2 min) The goal of database question answering is to enable natural language querying of real-life relational databases in diverse application domains. Recently, large-scale datasets such as Spider and WikiSQL facilitated novel modeling techniques for text-to-SQL parsing, improving zero-shot generalization to unseen databases. In this work, we examine the challenges that still prevent these techniques from practical deployment. First, we present KaggleDBQA, a new cross-domain evaluation dataset of real Web databases, with domain-specific data types, original formatting, and unrestricted questions. Second, we re-examine the choice of evaluation tasks for text-to-SQL parsers as applied in real-life settings. Finally, we augment our in-domain evaluation task with database documentation, a naturally occurring source of implicit domain knowledge. We show that KaggleDBQA presents a challenge to state-of-the-art zero-shot parsers but a more realistic evaluation setting and creative use of associated database documentation boosts their accuracy by over 13.2%, doubling their performance.

(2 min) Most modern neural networks for classification fail to take into account the concept of the unknown. Trained neural networks are usually tested in an unrealistic scenario with only examples from a closed set of known classes. In an attempt to develop a more realistic model, the concept of working in an open set environment has been introduced. This in turn leads to the concept of incremental learning where a model with its own architecture and initial trained set of data can identify unknown classes during the testing phase and autonomously update itself if evidence of a new class is detected. Some problems that arise in incremental learning are inefficient use of resources to retrain the classifier repeatedly and the decrease of classification accuracy as multiple classes are added over time. This process of instantiating new classes is repeated as many times as necessary, accruing errors. To address these problems, this paper proposes the Classification Confidence Threshold approach to prime neural networks for incremental learning to keep accuracies high by limiting forgetting. A lean method is also used to reduce resources used in the retraining of the neural network. The proposed method is based on the idea that a network is able to incrementally learn a new class even when exposed to a limited number samples associated with the new class. This method can be applied to most existing neural networks with minimal changes to network architecture.

(2 min) Despite inextricable ties between race and language, little work has considered race in NLP research and development. In this work, we survey 79 papers from the ACL anthology that mention race. These papers reveal various types of race-related bias in all stages of NLP model development, highlighting the need for proactive consideration of how NLP systems can uphold racial hierarchies. However, persistent gaps in research on race and NLP remain: race has been siloed as a niche topic and remains ignored in many NLP tasks; most work operationalizes race as a fixed single-dimensional variable with a ground-truth label, which risks reinforcing differences produced by historical racism; and the voices of historically marginalized people are nearly absent in NLP literature. By identifying where and how NLP literature has and has not considered race, especially in comparison to related fields, our work calls for inclusion and racial justice in NLP research practices.

(2 min) Neural machine translation (NMT) is sensitive to domain shift. In this paper, we address this problem in an active learning setting where we can spend a given budget on translating in-domain data, and gradually fine-tune a pre-trained out-of-domain NMT model on the newly translated data. Existing active learning methods for NMT usually select sentences based on uncertainty scores, but these methods require costly translation of full sentences even when only one or two key phrases within the sentence are informative. To address this limitation, we re-examine previous work from the phrase-based machine translation (PBMT) era that selected not full sentences, but rather individual phrases. However, while incorporating these phrases into PBMT systems was relatively simple, it is less trivial for NMT systems, which need to be trained on full sequences to capture larger structural properties of sentences unique to the new domain. To overcome these hurdles, we propose to select both full sentences and individual phrases from unlabelled data in the new domain for routing to human translators. In a German-English translation task, our active learning approach achieves consistent improvements over uncertainty-based sentence selection methods, improving up to 1.2 BLEU score over strong active learning baselines.

(2 min) Recently, the development of pre-trained language models has brought natural language processing (NLP) tasks to the new state-of-the-art. In this paper we explore the efficiency of various pre-trained language models. We pre-train a list of transformer-based models with the same amount of text and the same training steps. The experimental results shows that the most improvement upon the origin BERT is adding the RNN-layer to capture more contextual information for the transformer-encoder layers.

(2 min) Objective: The objective of this study is to develop a deep learning pipeline to detect signals on dietary supplement-related adverse events (DS AEs) from Twitter. Material and Methods: We obtained 247,807 tweets ranging from 2012 to 2018 that mentioned both DS and AE. We annotated biomedical entities and relations on 2,000 randomly selected tweets. For the concept extraction task, we compared the performance of traditional word embeddings with SVM, CRF and LSTM-CRF classifiers to BERT models. For the relation extraction task, we compared GloVe vectors with CNN classifiers to BERT models. We chose the best performing models in each task to assemble an end-to-end deep learning pipeline to detect DS AE signals and compared the results to the known DS AEs from a DS knowledge base (i.e., iDISK). Results: In both tasks, the BERT-based models outperformed traditional word embeddings. The best performing concept extraction model is the BioBERT model that can identify supplement, symptom, and body organ entities with F1-scores of 0.8646, 0.8497, and 0.7104, respectively. The best performing relation extraction model is the BERT model that can identify purpose and AE relations with F1-scores of 0.8335 and 0.7538, respectively. The end-to-end pipeline was able to extract DS indication and DS AEs with an F1-score of 0.7459 and 0,7414, respectively. Comparing to the iDISK, we could find both known and novel DS-AEs. Conclusion: We have demonstrated the feasibility of detecting DS AE signals from Twitter with a BioBERT-based deep learning pipeline.

(2 min) In this paper, we illustrate how to fine-tune the entire Retrieval Augment Generation (RAG) architecture in an end-to-end manner. We highlighted the main engineering challenges that needed to be addressed to achieve this objective. We also compare how end-to-end RAG architecture outperforms the original RAG architecture for the task of question answering. We have open-sourced our implementation in the HuggingFace Transformers library.

(2 min) This open-source book represents our attempt to make deep learning approachable, teaching readers the concepts, the context, and the code. The entire book is drafted in Jupyter notebooks, seamlessly integrating exposition figures, math, and interactive examples with self-contained code. Our goal is to offer a resource that could (i) be freely available for everyone; (ii) offer sufficient technical depth to provide a starting point on the path to actually becoming an applied machine learning scientist; (iii) include runnable code, showing readers how to solve problems in practice; (iv) allow for rapid updates, both by us and also by the community at large; (v) be complemented by a forum for interactive discussion of technical details and to answer questions.

(2 min) In the text processing context, most ML models are built on word embeddings. These embeddings are themselves trained on some datasets, potentially containing sensitive data. In some cases this training is done independently, in other cases, it occurs as part of training a larger, task-specific model. In either case, it is of interest to consider membership inference attacks based on the embedding layer as a way of understanding sensitive information leakage. But, somewhat surprisingly, membership inference attacks on word embeddings and their effect in other natural language processing (NLP) tasks that use these embeddings, have remained relatively unexplored. In this work, we show that word embeddings are vulnerable to black-box membership inference attacks under realistic assumptions. Furthermore, we show that this leakage persists through two other major NLP applications: classification and text-generation, even when the embedding layer is not exposed to the attacker. We show that our MI attack achieves high attack accuracy against a classifier model and an LSTM-based language model. Indeed, our attack is a cheaper membership inference attack on text-generative models, which does not require the knowledge of the target model or any expensive training of text-generative models as shadow models.

(2 min) Spiking Neural Networks (SNNs) have been attached great importance due to their biological plausibility and high energy-efficiency on neuromorphic chips. As these chips are usually resource-constrained, the compression of SNNs is thus crucial along the road of practical use of SNNs. Most existing methods directly apply pruning approaches in artificial neural networks (ANNs) to SNNs, which ignore the difference between ANNs and SNNs, thus limiting the performance of the pruned SNNs. Besides, these methods are only suitable for shallow SNNs. In this paper, inspired by synaptogenesis and synapse elimination in the neural system, we propose gradient rewiring (Grad R), a joint learning algorithm of connectivity and weight for SNNs, that enables us to seamlessly optimize network structure without retraining. Our key innovation is to redefine the gradient to a new synaptic parameter, allowing better exploration of network structures by taking full advantage of the competition between pruning and regrowth of connections. The experimental results show that the proposed method achieves minimal loss of SNNs' performance on MNIST and CIFAR-10 dataset so far. Moreover, it reaches a $\sim$3.5% accuracy loss under unprecedented 0.73% connectivity, which reveals remarkable structure refining capability in SNNs. Our work suggests that there exists extremely high redundancy in deep SNNs. Our codes are available at https://github.com/Yanqi-Chen/Gradient-Rewiring.

(2 min) Semi-supervised learning has been gaining attention as it allows for performing image analysis tasks such as classification with limited labeled data. Some popular algorithms using Generative Adversarial Networks (GANs) for semi-supervised classification share a single architecture for classification and discrimination. However, this may require a model to converge to a separate data distribution for each task, which may reduce overall performance. While progress in semi-supervised learning has been made, less addressed are small-scale, fully-supervised tasks where even unlabeled data is unavailable and unattainable. We therefore, propose a novel GAN model namely External Classifier GAN (EC-GAN), that utilizes GANs and semi-supervised algorithms to improve classification in fully-supervised regimes. Our method leverages a GAN to generate artificial data used to supplement supervised classification. More specifically, we attach an external classifier, hence the name EC-GAN, to the GAN's generator, as opposed to sharing an architecture with the discriminator. Our experiments demonstrate that EC-GAN's performance is comparable to the shared architecture method, far superior to the standard data augmentation and regularization-based approach, and effective on a small, realistic dataset.

(2 min) Probabilistic Face Embeddings (PFE) can improve face recognition performance in unconstrained scenarios by integrating data uncertainty into the feature representation. However, existing PFE methods tend to be over-confident in estimating uncertainty and is too slow to apply to large-scale face matching. This paper proposes a regularized probabilistic face embedding method to improve the robustness and speed of PFE. Specifically, the mutual likelihood score (MLS) metric used in PFE is simplified to speedup the matching of face feature pairs. Then, an output-constraint loss is proposed to penalize the variance of the uncertainty output, which can regularize the output of the neural network. In addition, an identification preserving loss is proposed to improve the discriminative of the MLS metric, and a multi-layer feature fusion module is proposed to improve the neural network's uncertainty estimation ability. Comprehensive experiments show that the proposed method can achieve comparable or better results in 9 benchmarks than the state-of-the-art methods, and can improve the performance of risk-controlled face recognition. The code of our work is publicly available in GitHub (https://github.com/KaenChan/ProbFace).

(2 min) Aiming at facilitating a real-world, ever-evolving and scalable autonomous driving system, we present a large-scale benchmark for standardizing the evaluation of different self-supervised and semi-supervised approaches by learning from raw data, which is the first and largest benchmark to date. Existing autonomous driving systems heavily rely on `perfect' visual perception models (e.g., detection) trained using extensive annotated data to ensure the safety. However, it is unrealistic to elaborately label instances of all scenarios and circumstances (e.g., night, extreme weather, cities) when deploying a robust autonomous driving system. Motivated by recent powerful advances of self-supervised and semi-supervised learning, a promising direction is to learn a robust detection model by collaboratively exploiting large-scale unlabeled data and few labeled data. Existing dataset (e.g., KITTI, Waymo) either provides only a small amount of data or covers limited domains with full annotation, hindering the exploration of large-scale pre-trained models. Here, we release a Large-Scale Object Detection benchmark for Autonomous driving, named as SODA10M, containing 10 million unlabeled images and 20K images labeled with 6 representative object categories. To improve diversity, the images are collected every ten seconds per frame within 32 different cities under different weather conditions, periods and location scenes. We provide extensive experiments and deep analyses of existing supervised state-of-the-art detection models, popular self-supervised and semi-supervised approaches, and some insights about how to develop future models. The data and more up-to-date information have been released at https://soda-2d.github.io.

(2 min) Pruning is an effective method to reduce the memory footprint and FLOPs associated with neural network models. However, existing structured-pruning methods often result in significant accuracy degradation for moderate pruning levels. To address this problem, we introduce a new Hessian Aware Pruning (HAP) method coupled with a Neural Implant approach that uses second-order sensitivity as a metric for structured pruning. The basic idea is to prune insensitive components and to use a Neural Implant for moderately sensitive components, instead of completely pruning them. For the latter approach, the moderately sensitive components are replaced with with a low rank implant that is smaller and less computationally expensive than the original component. We use the relative Hessian trace to measure sensitivity, as opposed to the magnitude based sensitivity metric commonly used in the literature. We test HAP for both computer vision tasks and natural language tasks, and we achieve new state-of-the-art results. Specifically, HAP achieves less than $0.1\%$/$0.5\%$ degradation on PreResNet29/ResNet50 (CIFAR-10/ImageNet) with more than 70\%/50\% of parameters pruned. Meanwhile, HAP also achieves significantly better performance (up to 0.8\% with 60\% of parameters pruned) as compared to gradient based method for head pruning on transformer-based models. The framework has been open sourced and available online.

(2 min) Medical imaging plays a pivotal role in diagnosis and treatment in clinical practice. Inspired by the significant progress in automatic image captioning, various deep learning (DL)-based architectures have been proposed for generating radiology reports for medical images. However, model uncertainty (i.e., model reliability/confidence on report generation) is still an under-explored problem. In this paper, we propose a novel method to explicitly quantify both the visual uncertainty and the textual uncertainty for the task of radiology report generation. Such multi-modal uncertainties can sufficiently capture the model confidence scores at both the report-level and the sentence-level, and thus they are further leveraged to weight the losses for achieving more comprehensive model optimization. Our experimental results have demonstrated that our proposed method for model uncertainty characterization and estimation can provide more reliable confidence scores for radiology report generation, and our proposed uncertainty-weighted losses can achieve more comprehensive model optimization and result in state-of-the-art performance on a public radiology report dataset.

(2 min) Multiple object tracking faces several challenges that may be alleviated with trajectory information. Knowing the posterior locations of an object helps disambiguating and solving situations such as occlusions, re-identification, and identity switching. In this work, we show that trajectory estimation can become a key factor for tracking, and present TrajE, a trajectory estimator based on recurrent mixture density networks, as a generic module that can be added to existing object trackers. To provide several trajectory hypotheses, our method uses beam search. Also, relying on the same estimated trajectory, we propose to reconstruct a track after an occlusion occurs. We integrate TrajE into two state of the art tracking algorithms, CenterTrack [63] and Tracktor [3]. Their respective performances in the MOTChallenge 2017 test set are boosted 6.3 and 0.3 points in MOTA score, and 1.8 and 3.1 in IDF1, setting a new state of the art for the CenterTrack+TrajE configuration

(2 min) Recent work attempts to improve semantic segmentation performance by exploring well-designed architectures on a target dataset. However, it remains challenging to build a unified system that simultaneously learns from various datasets due to the inherent distribution shift across different datasets. In this paper, we present a simple, flexible, and general method for semantic segmentation, termed Cross-Dataset Collaborative Learning (CDCL). Given multiple labeled datasets, we aim to improve the generalization and discrimination of feature representations on each dataset. Specifically, we first introduce a family of Dataset-Aware Blocks (DAB) as the fundamental computing units of the network, which help capture homogeneous representations and heterogeneous statistics across different datasets. Second, we propose a Dataset Alternation Training (DAT) mechanism to efficiently facilitate the optimization procedure. We conduct extensive evaluations on four diverse datasets, i.e., Cityscapes, BDD100K, CamVid, and COCO Stuff, with single-dataset and cross-dataset settings. Experimental results demonstrate our method consistently achieves notable improvements over prior single-dataset and cross-dataset training methods without introducing extra FLOPs. Particularly, with the same architecture of PSPNet (ResNet-18), our method outperforms the single-dataset baseline by 5.65\%, 6.57\%, and 5.79\% of mIoU on the validation sets of Cityscapes, BDD100K, CamVid, respectively. Code and models will be released.

(2 min) Reconstructing the shape and appearance of real-world objects using measured 2D images has been a long-standing problem in computer vision. In this paper, we introduce a new analysis-by-synthesis technique capable of producing high-quality reconstructions through robust coarse-to-fine optimization and physics-based differentiable rendering. Unlike most previous methods that handle geometry and reflectance largely separately, our method unifies the optimization of both by leveraging image gradients with respect to both object reflectance and geometry. To obtain physically accurate gradient estimates, we develop a new GPU-based Monte Carlo differentiable renderer leveraging recent advances in differentiable rendering theory to offer unbiased gradients while enjoying better performance than existing tools like PyTorch3D and redner. To further improve robustness, we utilize several shape and material priors as well as a coarse-to-fine optimization strategy to reconstruct geometry. We demonstrate that our technique can produce reconstructions with higher quality than previous methods such as COLMAP and Kinect Fusion.

(2 min) Predefined evenly-distributed class centroids (PEDCC) can be widely used in models and algorithms of pattern classification, such as CNN classifiers, classification autoencoders, clustering, and semi-supervised learning, etc. Its basic idea is to predefine the class centers, which are evenly-distributed on the unit hypersphere in feature space, to maximize the inter-class distance. The previous method of generating PEDCC uses an iterative algorithm based on a charge model, that is, the initial values of various centers (charge positions) are randomly set from the normal distribution, and the charge positions are updated iteratively with the help of the repulsive force between charges of the same polarity. The class centers generated by the algorithm will produce some errors with the theoretically evenly-distributed points, and the generation time will be longer. This paper takes advantage of regular polyhedron in high-dimensional space and the evenly distribution of points on the n dimensional hypersphere to generate PEDCC mathematically. Then, we discussed the basic and extensive characteristics of the frames formed by PEDCC. Finally, experiments show that new algorithm is not only faster than the iterative method, but also more accurate in position. The mathematical analysis and experimental results of this paper can provide a theoretical tool for using PEDCC to solve the key problems in the field of pattern recognition, such as interpretable supervised/unsupervised learning, incremental learning, uncertainty analysis and so on.

(2 min) Deep neural networks designed for vision tasks are often prone to failure when they encounter environmental conditions not covered by the training data. Single-modal strategies are insufficient when the sensor fails to acquire information due to malfunction or its design limitations. Multi-sensor configurations are known to provide redundancy, increase reliability, and are crucial in achieving robustness against asymmetric sensor failures. To address the issue of changing lighting conditions and asymmetric sensor degradation in object detection, we develop a multi-modal 2D object detector, and propose deterministic and stochastic sensor-aware feature fusion strategies. The proposed fusion mechanisms are driven by the estimated sensor measurement reliability values/weights. Reliable object detection in harsh lighting conditions is essential for applications such as self-driving vehicles and human-robot interaction. We also propose a new "r-blended" hybrid depth modality for RGB-D sensors. Through extensive experimentation, we show that the proposed strategies outperform the existing state-of-the-art methods on the FLIR-Thermal dataset, and obtain promising results on the SUNRGB-D dataset. We additionally record a new RGB-Infra indoor dataset, namely L515-Indoors, and demonstrate that the proposed object detection methodologies are highly effective for a variety of lighting conditions.

(2 min) Background: Colonoscopy remains the gold-standard screening for colorectal cancer. However, significant miss rates for polyps have been reported, particularly when there are multiple small adenomas. This presents an opportunity to leverage computer-aided systems to support clinicians and reduce the number of polyps missed. Method: In this work we introduce the Focus U-Net, a novel dual attention-gated deep neural network, which combines efficient spatial and channel-based attention into a single Focus Gate module to encourage selective learning of polyp features. The Focus U-Net further incorporates short-range skip connections and deep supervision. Furthermore, we introduce the Hybrid Focal loss, a new compound loss function based on the Focal loss and Focal Tversky loss, to handle class-imbalanced image segmentation. For our experiments, we selected five public datasets containing images of polyps obtained during optical colonoscopy: CVC-ClinicDB, Kvasir-SEG, CVC-ColonDB, ETIS-Larib PolypDB and EndoScene test set. To evaluate model performance, we use the Dice similarity coefficient (DSC) and Intersection over Union (IoU) metrics. Results: Our model achieves state-of-the-art results for both CVC-ClinicDB and Kvasir-SEG, with a mean DSC of 0.941 and 0.910, respectively. When evaluated on a combination of five public polyp datasets, our model similarly achieves state-of-the-art results with a mean DSC of 0.878 and mean IoU of 0.809, a 14% and 15% improvement over the previous state-of-the-art results of 0.768 and 0.702, respectively. Conclusions: This study shows the potential for deep learning to provide fast and accurate polyp segmentation results for use during colonoscopy. The Focus U-Net may be adapted for future use in newer non-invasive screening and more broadly to other biomedical image segmentation tasks involving class imbalance and requiring efficiency.

(2 min) Recently demonstrated physical-world adversarial attacks have exposed vulnerabilities in perception systems that pose severe risks for safety-critical applications such as autonomous driving. These attacks place adversarial artifacts in the physical world that indirectly cause the addition of a universal patch to inputs of a model that can fool it in a variety of contexts. Adversarial training is the most effective defense against image-dependent adversarial attacks. However, tailoring adversarial training to universal patches is computationally expensive since the optimal universal patch depends on the model weights which change during training. We propose meta adversarial training (MAT), a novel combination of adversarial training with meta-learning, which overcomes this challenge by meta-learning universal patches along with model training. MAT requires little extra computation while continuously adapting a large set of patches to the current model. MAT considerably increases robustness against universal patch attacks on image classification and traffic-light detection.

(2 min) Self-supervised learning (especially contrastive learning) has attracted great interest due to its tremendous potentials in learning discriminative representations in an unsupervised manner. Despite the acknowledged successes, existing contrastive learning methods suffer from very low learning efficiency, e.g., taking about ten times more training epochs than supervised learning for comparable recognition accuracy. In this paper, we discover two contradictory phenomena in contrastive learning that we call under-clustering and over-clustering problems, which are major obstacles to learning efficiency. Under-clustering means that the model cannot efficiently learn to discover the dissimilarity between inter-class samples when the negative sample pairs for contrastive learning are insufficient to differentiate all the actual object categories. Over-clustering implies that the model cannot efficiently learn the feature representation from excessive negative sample pairs, which enforces the model to over-cluster samples of the same actual categories into different clusters. To simultaneously overcome these two problems, we propose a novel self-supervised learning framework using a median triplet loss. Precisely, we employ a triplet loss tending to maximize the relative distance between the positive pair and negative pairs to address the under-clustering problem; and we construct the negative pair by selecting the negative sample of a median similarity score from all negative samples to avoid the over-clustering problem, guaranteed by the Bernoulli Distribution model. We extensively evaluate our proposed framework in several large-scale benchmarks (e.g., ImageNet, SYSU-30k, and COCO). The results demonstrate the superior performance (e.g., the learning efficiency) of our model over the latest state-of-the-art methods by a clear margin. Codes available at: https://github.com/wanggrun/triplet.

(2 min) Compressive lensless imagers enable novel applications in an extremely compact device, requiring only a phase or amplitude mask placed close to the sensor. They have been demonstrated for 2D and 3D microscopy, single-shot video, and single-shot hyperspectral imaging; in each of these cases, a compressive-sensing-based inverse problem is solved in order to recover a 3D data-cube from a 2D measurement. Typically, this is accomplished using convex optimization and hand-picked priors. Alternatively, deep learning-based reconstruction methods offer the promise of better priors, but require many thousands of ground truth training pairs, which can be difficult or impossible to acquire. In this work, we propose the use of untrained networks for compressive image recovery. Our approach does not require any labeled training data, but instead uses the measurement itself to update the network weights. We demonstrate our untrained approach on lensless compressive 2D imaging as well as single-shot high-speed video recovery using the camera's rolling shutter, and single-shot hyperspectral imaging. We provide simulation and experimental verification, showing that our method results in improved image quality over existing methods.

(2 min) In this work, we present a novel neural network to generate high resolution images. We replace the decoder of VAE with a discriminator while using the encoder as it is. The encoder is fed data from a normal distribution while the generator is fed from a gaussian distribution. The combination from both is given to a discriminator which tells whether the generated image is correct or not. We evaluate our network on 3 different datasets: MNIST, LSUN and CelebA dataset. Our network beats the previous state of the art using MMD, SSIM, log likelihood, reconstruction error, ELBO and KL divergence as the evaluation metrics while generating much sharper images. This work is potentially very exciting as we are able to combine the advantages of generative models and inference models in a principled bayesian manner.

(2 min) Contrastive self-supervised learning has largely narrowed the gap to supervised pre-training on ImageNet. However, its success highly relies on the object-centric priors of ImageNet, i.e., different augmented views of the same image correspond to the same object. Such a heavily curated constraint becomes immediately infeasible when pre-trained on more complex scene images with many objects. To overcome this limitation, we introduce Object-level Representation Learning (ORL), a new self-supervised learning framework towards scene images. Our key insight is to leverage image-level self-supervised pre-training as the prior to discover object-level semantic correspondence, thus realizing object-level representation learning from scene images. Extensive experiments on COCO show that ORL significantly improves the performance of self-supervised learning on scene images, even surpassing supervised ImageNet pre-training on several downstream tasks. Furthermore, ORL improves the downstream performance when more unlabeled scene images are available, demonstrating its great potential of harnessing unlabeled data in the wild. We hope our approach can motivate future research on more general-purpose unsupervised representation learning from scene data. Project page: https://www.mmlab-ntu.com/project/orl/.

(2 min) In large-scale optimization, the presence of nonsmooth and nonconvex terms in a given problem typically makes it hard to solve. A popular approach to address nonsmooth terms in convex optimization is to approximate them with their respective Moreau envelopes. In this work, we study the use of Lasry-Lions double envelopes to approximate nonsmooth terms that are also not convex. These envelopes are an extension of the Moreau ones but exhibit an additional smoothness property that makes them amenable to fast optimization algorithms. Lasry-Lions envelopes can also be seen as an "intermediate" between a given function and its convex envelope, and we make use of this property to develop a method that builds a sequence of approximate subproblems that are easier to solve than the original problem. We discuss convergence properties of this method when used to address composite minimization problems; additionally, based on a number of experiments, we discuss settings where it may be more useful than classical alternatives in two domains: signal decoding and spectral unmixing.

(2 min) Deep hashing models have been proposed as an efficient method for large-scale similarity search. However, most existing deep hashing methods only utilize fine-level labels for training while ignoring the natural semantic hierarchy structure. This paper presents an effective method that preserves the classwise similarity of full-level semantic hierarchy for large-scale image retrieval. Experiments on two benchmark datasets show that our method helps improve the fine-level retrieval performance. Moreover, with the help of the semantic hierarchy, it can produce significantly better binary codes for hierarchical retrieval, which indicates its potential of providing more user-desired retrieval results.

(2 min) Recently, the single image super-resolution (SISR) approaches with deep and complex convolutional neural network structures have achieved promising performance. However, those methods improve the performance at the cost of higher memory consumption, which is difficult to be applied for some mobile devices with limited storage and computing resources. To solve this problem, we present a lightweight multi-scale feature interaction network (MSFIN). For lightweight SISR, MSFIN expands the receptive field and adequately exploits the informative features of the low-resolution observed images from various scales and interactive connections. In addition, we design a lightweight recurrent residual channel attention block (RRCAB) so that the network can benefit from the channel attention mechanism while being sufficiently lightweight. Extensive experiments on some benchmarks have confirmed that our proposed MSFIN can achieve comparable performance against the state-of-the-arts with a more lightweight model.

(3 min) As soon as abstract mathematical computations were adapted to computation on digital computers, the problem of efficient representation, manipulation, and communication of the numerical values in those computations arose. Strongly related to the problem of numerical representation is the problem of quantization: in what manner should a set of continuous real-valued numbers be distributed over a fixed discrete set of numbers to minimize the number of bits required and also to maximize the accuracy of the attendant computations? This perennial problem of quantization is particularly relevant whenever memory and/or computational resources are severely restricted, and it has come to the forefront in recent years due to the remarkable performance of Neural Network models in computer vision, natural language processing, and related areas. Moving from floating-point representations to low-precision fixed integer values represented in four bits or less holds the potential to reduce the memory footprint and latency by a factor of 16x; and, in fact, reductions of 4x to 8x are often realized in practice in these applications. Thus, it is not surprising that quantization has emerged recently as an important and very active sub-area of research in the efficient implementation of computations associated with Neural Networks. In this article, we survey approaches to the problem of quantizing the numerical values in deep Neural Network computations, covering the advantages/disadvantages of current methods. With this survey and its organization, we hope to have presented a useful snapshot of the current research in quantization for Neural Networks and to have given an intelligent organization to ease the evaluation of future research in this area.

(2 min) Conventional image classifiers are trained by randomly sampling mini-batches of images. To achieve state-of-the-art performance, practitioners use sophisticated data augmentation schemes to expand the amount of training data available for sampling. In contrast, meta-learning algorithms sample support data, query data, and tasks on each training step. In this complex sampling scenario, data augmentation can be used not only to expand the number of images available per class, but also to generate entirely new classes/tasks. We systematically dissect the meta-learning pipeline and investigate the distinct ways in which data augmentation can be integrated at both the image and class levels. Our proposed meta-specific data augmentation significantly improves the performance of meta-learners on few-shot classification benchmarks.

(2 min) With the introduction of new regulations in the European Union, the future of Beyond Visual Line Of Sight (BVLOS) drones is set to bloom. This led to the creation of the theBEAST project, which aims to create an autonomous security drone, with focus on those regulations and on safety. This technical paper describes the first steps of a module within this project, which revolves around detecting obstacles so they can be avoided in a fail-safe landing. A deep learning powered object detection method is the subject of our research, and various experiments are held to maximize its performance, such as comparing various data augmentation techniques or YOLOv3 and YOLOv5. According to the results of the experiments, we conclude that although object detection is a promising approach to resolve this problem, more volume of data is required for potential usage in a real-life application.

(2 min) Few-shot learning aims to transfer information from one task to enable generalization on novel tasks given a few examples. This information is present both in the domain and the class labels. In this work we investigate the complementary roles of these two sources of information by combining instance-discriminative contrastive learning and supervised learning in a single framework called Supervised Momentum Contrastive learning (SUPMOCO). Our approach avoids a problem observed in supervised learning where information in images not relevant to the task is discarded, which hampers their generalization to novel tasks. We show that (self-supervised) contrastive learning and supervised learning are mutually beneficial, leading to a new state-of-the-art on the META-DATASET - a recently introduced benchmark for few-shot learning. Our method is based on a simple modification of MOCO and scales better than prior work on combining supervised and self-supervised learning. This allows us to easily combine data from multiple domains leading to further improvements.

(2 min) In this work, we address the problem of 3D object detection from point cloud data in real time. For autonomous vehicles to work, it is very important for the perception component to detect the real world objects with both high accuracy and fast inference. We propose a novel neural network architecture along with the training and optimization details for detecting 3D objects in point cloud data. We compare the results with different backbone architectures including the standard ones like VGG, ResNet, Inception with our backbone. Also we present the optimization and ablation studies including designing an efficient anchor. We use the Kitti 3D Birds Eye View dataset for benchmarking and validating our results. Our work surpasses the state of the art in this domain both in terms of average precision and speed running at > 30 FPS. This makes it a feasible option to be deployed in real time applications including self driving cars.

(2 min) In class-incremental learning, an agent with limited resources needs to learn a sequence of classification tasks, forming an ever growing classification problem, with the constraint of not being able to access data from previous tasks. The main difference with task-incremental learning, where a task-ID is available at inference time, is that the learner also needs to perform cross-task discrimination, i.e. distinguish between classes that have not been seen together. Approaches to tackle this problem are numerous and mostly make use of an external memory (buffer) of non-negligible size. In this paper, we ablate the learning of cross-task features and study its influence on the performance of basic replay strategies used for class-IL. We also define a new forgetting measure for class-incremental learning, and see that forgetting is not the principal cause of low performance. Our experimental results show that future algorithms for class-incremental learning should not only prevent forgetting, but also aim to improve the quality of the cross-task features. This is especially important when the number of classes per task is small.

(2 min) In this paper, we aim to create generalizable and controllable neural signed distance fields (SDFs) that represent clothed humans from monocular depth observations. Recent advances in deep learning, especially neural implicit representations, have enabled human shape reconstruction and controllable avatar generation from different sensor inputs. However, to generate realistic cloth deformations from novel input poses, watertight meshes or dense full-body scans are usually needed as inputs. Furthermore, due to the difficulty of effectively modeling pose-dependent cloth deformations for diverse body shapes and cloth types, existing approaches resort to per-subject/cloth-type optimization from scratch, which is computationally expensive. In contrast, we propose an approach that can quickly generate realistic clothed human avatars, represented as controllable neural SDFs, given only monocular depth images. We achieve this by using meta-learning to learn an initialization of a hypernetwork that predicts the parameters of neural SDFs. The hypernetwork is conditioned on human poses and represents a clothed neural avatar that deforms non-rigidly according to the input poses. Meanwhile, it is meta-learned to effectively incorporate priors of diverse body shapes and cloth types and thus can be much faster to fine-tune, compared to models trained from scratch. We qualitatively and quantitatively show that our approach outperforms state-of-the-art approaches that require complete meshes as inputs while our approach requires only depth frames as inputs and runs orders of magnitudes faster. Furthermore, we demonstrate that our meta-learned hypernetwork is very robust, being the first to generate avatars with realistic dynamic cloth deformations given as few as 8 monocular depth frames.

(2 min) Skeleton-based Human Activity Recognition has achieved great interest in recent years as skeleton data has demonstrated being robust to illumination changes, body scales, dynamic camera views, and complex background. In particular, Spatial-Temporal Graph Convolutional Networks (ST-GCN) demonstrated to be effective in learning both spatial and temporal dependencies on non-Euclidean data such as skeleton graphs. Nevertheless, an effective encoding of the latent information underlying the 3D skeleton is still an open problem, especially when it comes to extracting effective information from joint motion patterns and their correlations. In this work, we propose a novel Spatial-Temporal Transformer network (ST-TR) which models dependencies between joints using the Transformer self-attention operator. In our ST-TR model, a Spatial Self-Attention module (SSA) is used to understand intra-frame interactions between different body parts, and a Temporal Self-Attention module (TSA) to model inter-frame correlations. The two are combined in a two-stream network, whose performance is evaluated on three large-scale datasets, NTU-RGB+D 60, NTU-RGB+D 120, and Kinetics Skeleton 400, consistently improving backbone results. Compared with methods that use the same input data, the proposed ST-TR achieves state-of-the-art performance on all datasets when using joints' coordinates as input, and results on-par with state-of-the-art when adding bones information.

(2 min) Deep neural networks have reached very high accuracy on object detection but their success hinges on large amounts of labeled data. To reduce the dependency on labels, various active-learning strategies have been proposed, typically based on the confidence of the detector. However, these methods are biased towards best-performing classes and can lead to acquired datasets that are not good representatives of the data in the testing set. In this work, we propose a unified framework for active learning, that considers both the uncertainty and the robustness of the detector, ensuring that the network performs accurately in all classes. Furthermore, our method is able to pseudo-label the very confident predictions, suppressing a potential distribution drift while further boosting the performance of the model. Experiments show that our method comprehensively outperforms a wide range of active-learning methods on PASCAL VOC07+12 and MS-COCO, having up to a 7.7% relative improvement, or up to 82% reduction in labeling cost.

(2 min) Understanding the nutritional content of food from visual data is a challenging computer vision problem, with the potential to have a positive and widespread impact on public health. Studies in this area are limited to existing datasets in the field that lack sufficient diversity or labels required for training models with nutritional understanding capability. We introduce Nutrition5k, a novel dataset of 5k diverse, real world food dishes with corresponding video streams, depth images, component weights, and high accuracy nutritional content annotation. We demonstrate the potential of this dataset by training a computer vision algorithm capable of predicting the caloric and macronutrient values of a complex, real world dish at an accuracy that outperforms professional nutritionists. Further we present a baseline for incorporating depth sensor data to improve nutrition predictions. We will publicly release Nutrition5k in the hope that it will accelerate innovation in the space of nutritional understanding.

(2 min) We present an algorithm for estimating consistent dense depth maps and camera poses from a monocular video. We integrate a learning-based depth prior, in the form of a convolutional neural network trained for single-image depth estimation, with geometric optimization, to estimate a smooth camera trajectory as well as detailed and stable depth reconstruction. Our algorithm combines two complementary techniques: (1) flexible deformation-splines for low-frequency large-scale alignment and (2) geometry-aware depth filtering for high-frequency alignment of fine depth details. In contrast to prior approaches, our method does not require camera poses as input and achieves robust reconstruction for challenging hand-held cell phone captures containing a significant amount of noise, shake, motion blur, and rolling shutter deformations. Our method quantitatively outperforms state-of-the-arts on the Sintel benchmark for both depth and pose estimations and attains favorable qualitative results across diverse wild datasets.

(2 min) Voice anti-spoofing aims at classifying a given utterance either as a bonafide human sample, or a spoofing attack (e.g. synthetic or replayed sample). Many anti-spoofing methods have been proposed but most of them fail to generalize across domains (corpora) -- and we do not know \emph{why}. We outline a novel interpretative framework for gauging the impact of data quality upon anti-spoofing performance. Our within- and between-domain experiments pool data from seven public corpora and three anti-spoofing methods based on Gaussian mixture and convolutive neural network models. We assess the impacts of long-term spectral information, speaker population (through x-vector speaker embeddings), signal-to-noise ratio, and selected voice quality features.

(2 min) Image segmentation is a common and challenging task in autonomous driving. Availability of sufficient pixel-level annotations for the training data is a hurdle. Active learning helps learning from small amounts of data by suggesting the most promising samples for labeling. In this work, we propose a new pool-based method for active learning, which proposes promising image regions, in each acquisition step. The problem is framed in an exploration-exploitation framework by combining an embedding based on Uniform Manifold Approximation to model representativeness with entropy as uncertainty measure to model informativeness. We applied our proposed method to the challenging autonomous driving data sets CamVid and Cityscapes and performed a quantitative comparison with state-of-the-art methods. We find that our active learning method achieves better performance on CamVid compared to other methods, while on Cityscapes, the performance lift was negligible.

(2 min) Analyzing the structure of proteins is a key part of understanding their functions and thus their role in biology at the molecular level. In addition, design new proteins in a methodical way is a major engineering challenge. In this work, we introduce a joint geometric-neural networks approach for comparing, deforming and generating 3D protein structures. Viewing protein structures as 3D open curves, we adopt the Square Root Velocity Function (SRVF) representation and leverage its suitable geometric properties along with Deep Residual Networks (ResNets) for a joint registration and comparison. Our ResNets handle better large protein deformations while being more computationally efficient. On top of the mathematical framework, we further design a Geometric Variational Auto-Encoder (G-VAE), that once trained, maps original, previously unseen structures, into a low-dimensional (latent) hyper-sphere. Motivated by the spherical structure of the pre-shape space, we naturally adopt the von Mises-Fisher (vMF) distribution to model our hidden variables. We test the effectiveness of our models by generating novel protein structures and predicting completions of corrupted protein structures. Experimental results show that our method is able to generate plausible structures, different from the structures in the training data.

(2 min) In this work, we propose the world's first closed-loop ML-based planning benchmark for autonomous driving. While there is a growing body of ML-based motion planners, the lack of established datasets and metrics has limited the progress in this area. Existing benchmarks for autonomous vehicle motion prediction have focused on short-term motion forecasting, rather than long-term planning. This has led previous works to use open-loop evaluation with L2-based metrics, which are not suitable for fairly evaluating long-term planning. Our benchmark overcomes these limitations by introducing a large-scale driving dataset, lightweight closed-loop simulator, and motion-planning-specific metrics. We provide a high-quality dataset with 1500h of human driving data from 4 cities across the US and Asia with widely varying traffic patterns (Boston, Pittsburgh, Las Vegas and Singapore). We will provide a closed-loop simulation framework with reactive agents and provide a large set of both general and scenario-specific planning metrics. We plan to release the dataset at NeurIPS 2021 and organize benchmark challenges starting in early 2022.

(2 min) It is well known that deep neural networks (DNNs) are vulnerable to adversarial attacks, which are implemented by adding crafted perturbations onto benign examples. Min-max robust optimization based adversarial training can provide a notion of security against adversarial attacks. However, adversarial robustness requires a significantly larger capacity of the network than that for the natural training with only benign examples. This paper proposes a framework of concurrent adversarial training and weight pruning that enables model compression while still preserving the adversarial robustness and essentially tackles the dilemma of adversarial training. Furthermore, this work studies two hypotheses about weight pruning in the conventional setting and finds that weight pruning is essential for reducing the network model size in the adversarial setting, training a small model from scratch even with inherited initialization from the large model cannot achieve both adversarial robustness and high standard accuracy. Code is available at https://github.com/yeshaokai/Robustness-Aware-Pruning-ADMM.

(2 min) Speech production is a dynamic procedure, which involved multi human organs including the tongue, jaw and lips. Modeling the dynamics of the vocal tract deformation is a fundamental problem to understand the speech, which is the most common way for human daily communication. Researchers employ several sensory streams to describe the process simultaneously, which are incontrovertibly statistically related to other streams. In this paper, we address the following question: given an observable image sequences of lips, can we picture the corresponding tongue motion. We formulated this problem as the self-supervised learning problem, and employ the two-stream convolutional network and long-short memory network for the learning task, with the attention mechanism. We evaluate the performance of the proposed method by leveraging the unlabeled lip videos to predict an upcoming ultrasound tongue image sequence. The results show that our model is able to generate images that close to the real ultrasound tongue images, and results in the matching between two imaging modalities.

(2 min) Multiple object tracking and segmentation requires detecting, tracking, and segmenting objects belonging to a set of given classes. Most approaches only exploit the temporal dimension to address the association problem, while relying on single frame predictions for the segmentation mask itself. We propose Prototypical Cross-Attention Network (PCAN), capable of leveraging rich spatio-temporal information for online multiple object tracking and segmentation. PCAN first distills a space-time memory into a set of prototypes and then employs cross-attention to retrieve rich information from the past frames. To segment each object, PCAN adopts a prototypical appearance module to learn a set of contrastive foreground and background prototypes, which are then propagated over time. Extensive experiments demonstrate that PCAN outperforms current video instance tracking and segmentation competition winners on both Youtube-VIS and BDD100K datasets, and shows efficacy to both one-stage and two-stage segmentation frameworks. Code will be available at this http URL

(2 min) Organ-at-risk contouring is still a bottleneck in radiotherapy, with many deep learning methods falling short of promised results when evaluated on clinical data. We investigate the accuracy and time-savings resulting from the use of an interactive-machine-learning method for an organ-at-risk contouring task. We compare the method to the Eclipse contouring software and find strong agreement with manual delineations, with a dice score of 0.95. The annotations created using corrective-annotation also take less time to create as more images are annotated, resulting in substantial time savings compared to manual methods, with hearts that take 2 minutes and 2 seconds to delineate on average, after 923 images have been delineated, compared to 7 minutes and 1 seconds when delineating manually. Our experiment demonstrates that interactive-machine-learning with corrective-annotation provides a fast and accessible way for non computer-scientists to train deep-learning models to segment their own structures of interest as part of routine clinical workflows. Source code is available at \href{https://github.com/Abe404/RootPainter3D}{this HTTPS URL}.

(2 min) Data augmentation has been successfully used in many areas of deep-learning to significantly improve model performance. Typically data augmentation simulates realistic variations in data in order to increase the apparent diversity of the training-set. However, for opcode-based malware analysis, where deep learning methods are already achieving state of the art performance, it is not immediately clear how to apply data augmentation. In this paper we study different methods of data augmentation starting with basic methods using fixed transformations and moving to methods that adapt to the data. We propose a novel data augmentation method based on using an opcode embedding layer within the network and its corresponding opcode embedding matrix to perform adaptive data augmentation during training. To the best of our knowledge this is the first paper to carry out a systematic study of different augmentation methods applied to opcode sequence based malware classification.

(2 min) We propose a method to detect and reconstruct multiple 3D objects from a single RGB image. The key idea is to optimize for detection, alignment and shape jointly over all objects in the RGB image, while focusing on realistic and physically plausible reconstructions. To this end, we propose a keypoint detector that localizes objects as center points and directly predicts all object properties, including 9-DoF bounding boxes and 3D shapes -- all in a single forward pass. The proposed method formulates 3D shape reconstruction as a shape selection problem, i.e. it selects among exemplar shapes from a given database. This makes it agnostic to shape representations, which enables a lightweight reconstruction of realistic and visually-pleasing shapes based on CAD-models, while the training objective is formulated around point clouds and voxel representations. A collision-loss promotes non-intersecting objects, further increasing the reconstruction realism. Given the RGB image, the presented approach performs lightweight reconstruction in a single-stage, it is real-time capable, fully differentiable and end-to-end trainable. Our experiments compare multiple approaches for 9-DoF bounding box estimation, evaluate the novel shape-selection mechanism and compare to recent methods in terms of 3D bounding box estimation and 3D shape reconstruction quality.

(2 min) Recently, query based deep networks catch lots of attention owing to their end-to-end pipeline and competitive results on several fundamental computer vision tasks, such as object detection, semantic segmentation, and instance segmentation. However, how to establish a query based video instance segmentation (VIS) framework with elegant architecture and strong performance remains to be settled. In this paper, we present \textbf{QueryTrack} (i.e., tracking instances as queries), a unified query based VIS framework fully leveraging the intrinsic one-to-one correspondence between instances and queries in QueryInst. The proposed method obtains 52.7 / 52.3 AP on YouTube-VIS-2019 / 2021 datasets, which wins the 2-nd place in the YouTube-VIS Challenge at CVPR 2021 \textbf{with a single online end-to-end model, single scale testing \& modest amount of training data}. We also provide QueryTrack-ResNet-50 baseline results on YouTube-VIS-2021 dataset as references for the VIS community.

(2 min) With the advancement of deep neural networks and computer vision-based Human Activity Recognition, employment of Point-Cloud Data technologies (LiDAR, mmWave) has seen a lot interests due to its privacy preserving nature. Given the high promise of accurate PCD technologies, we develop, PALMAR, a multiple-inhabitant activity recognition system by employing efficient signal processing and novel machine learning techniques to track individual person towards developing an adaptive multi-inhabitant tracking and HAR system. More specifically, we propose (i) a voxelized feature representation-based real-time PCD fine-tuning method, (ii) efficient clustering (DBSCAN and BIRCH), Adaptive Order Hidden Markov Model based multi-person tracking and crossover ambiguity reduction techniques and (iii) novel adaptive deep learning-based domain adaptation technique to improve the accuracy of HAR in presence of data scarcity and diversity (device, location and population diversity). We experimentally evaluate our framework and systems using (i) a real-time PCD collected by three devices (3D LiDAR and 79 GHz mmWave) from 6 participants, (ii) one publicly available 3D LiDAR activity data (28 participants) and (iii) an embedded hardware prototype system which provided promising HAR performances in multi-inhabitants (96%) scenario with a 63% improvement of multi-person tracking than state-of-art framework without losing significant system performances in the edge computing device.

(2 min) Dietary-related problems such as obesity are a growing concern in todays modern world. If the current trend continues, it is most likely that the quality of life, in general, is significantly affected since obesity is associated with other chronic diseases such as hypertension, irregular blood sugar levels, and increased risk of heart attacks. The primary cause of these problems is poor lifestyle choices and unhealthy dietary habits, with emphasis on a select few food groups such as sugars, fats, and carbohydrates. In this regard, computer-based food recognition offers automatic visual-based methods to assess dietary intake and help people make healthier choices. Thus, the following paper presents a brief review of visual-based methods for food recognition, including their accuracy, performance, and the use of popular food databases to evaluate existing models. The work further aims to highlight future challenges in this area. New high-quality studies for developing standard benchmarks and using continual learning methods for food recognition are recommended.

(2 min) Nowadays, analysis of transparent images in the field of computer vision has gradually become a hot spot. In this paper, we compare the classification performance of different deep learning for the problem that transparent images are difficult to analyze. We crop the transparent images into 8 * 8 and 224 * 224 pixels patches in the same proportion, and then divide the two different pixels patches into foreground and background according to groundtruch. We also use 4 types of convolutional neural networks and a novel ViT network model to compare the foreground and background classification experiments. We conclude that ViT performs the worst in classifying 8 * 8 pixels patches, but it outperforms most convolutional neural networks in classifying 224 * 224.

(2 min) Domain adaptation aims to mitigate the domain gap when transferring knowledge from an existing labeled domain to a new domain. However, existing disentanglement-based methods do not fully consider separation between domain-invariant and domain-specific features, which means the domain-invariant features are not discriminative. The reconstructed features are also not sufficiently used during training. In this paper, we propose a novel enhanced separable disentanglement (ESD) model. We first employ a disentangler to distill domain-invariant and domain-specific features. Then, we apply feature separation enhancement processes to minimize contamination between domain-invariant and domain-specific features. Finally, our model reconstructs complete feature vectors, which are used for further disentanglement during the training phase. Extensive experiments from three benchmark datasets outperform state-of-the-art methods, especially on challenging cross-domain tasks.

(2 min) This paper introduces an approach for multi-human 3D pose estimation and tracking based on calibrated multi-view. The main challenge lies in finding the cross-view and temporal correspondences correctly even when several human pose estimations are noisy. Compare to previous solutions that construct 3D poses from multiple views, our approach takes advantage of temporal consistency to match the 2D poses estimated with previously constructed 3D skeletons in every view. Therefore cross-view and temporal associations are accomplished simultaneously. Since the performance suffers from mistaken association and noisy predictions, we design two strategies for aiming better correspondences and 3D reconstruction. Specifically, we propose a part-aware measurement for 2D-3D association and a filter that can cope with 2D outliers during reconstruction. Our approach is efficient and effective comparing to state-of-the-art methods; it achieves competitive results on two benchmarks: 96.8% on Campus and 97.4% on Shelf. Moreover, we extends the length of Campus evaluation frames to be more challenging and our proposal also reach well-performed result.

(2 min) We present HybVIO, a novel hybrid approach for combining filtering-based visual-inertial odometry (VIO) with optimization-based SLAM. The core of our method is highly robust, independent VIO with improved IMU bias modeling, outlier rejection, stationarity detection, and feature track selection, which is adjustable to run on embedded hardware. Long-term consistency is achieved with a loosely-coupled SLAM module. In academic benchmarks, our solution yields excellent performance in all categories, especially in the real-time use case, where we outperform the current state-of-the-art. We also demonstrate the feasibility of VIO for vehicular tracking on consumer-grade hardware using a custom dataset, and show good performance in comparison to current commercial VISLAM alternatives.

(2 min) Weakly-Supervised Temporal Action Localization (WS-TAL) task aims to recognize and localize temporal starts and ends of action instances in an untrimmed video with only video-level label supervision. Due to lack of negative samples of background category, it is difficult for the network to separate foreground and background, resulting in poor detection performance. In this report, we present our 2021 HACS Challenge - Weakly-supervised Learning Track solution that based on BaSNet to address above problem. Specifically, we first adopt pre-trained CSN, Slowfast, TDN, and ViViT as feature extractors to get feature sequences. Then our proposed Local-Global Background Modeling Network (LGBM-Net) is trained to localize instances by using only video-level labels based on Multi-Instance Learning (MIL). Finally, we ensemble multiple models to get the final detection results and reach 22.45% mAP on the test set

(2 min) Chinese character recognition has attracted much research interest due to its wide applications. Although it has been studied for many years, some issues in this field have not been completely resolved yet, e.g. the zero-shot problem. Previous character-based and radical-based methods have not fundamentally addressed the zero-shot problem since some characters or radicals in test sets may not appear in training sets under a data-hungry condition. Inspired by the fact that humans can generalize to know how to write characters unseen before if they have learned stroke orders of some characters, we propose a stroke-based method by decomposing each character into a sequence of strokes, which are the most basic units of Chinese characters. However, we observe that there is a one-to-many relationship between stroke sequences and Chinese characters. To tackle this challenge, we employ a matching-based strategy to transform the predicted stroke sequence to a specific character. We evaluate the proposed method on handwritten characters, printed artistic characters, and scene characters. The experimental results validate that the proposed method outperforms existing methods on both character zero-shot and radical zero-shot tasks. Moreover, the proposed method can be easily generalized to other languages whose characters can be decomposed into strokes.

(2 min) Zero-Shot Sketch-Based Image Retrieval (ZS-SBIR) is a novel cross-modal retrieval task, where abstract sketches are used as queries to retrieve natural images under zero-shot scenario. Most existing methods regard ZS-SBIR as a traditional classification problem and employ a cross-entropy or triplet-based loss to achieve retrieval, which neglect the problems of the domain gap between sketches and natural images and the large intra-class diversity in sketches. Toward this end, we propose a novel Domain-Smoothing Network (DSN) for ZS-SBIR. Specifically, a cross-modal contrastive method is proposed to learn generalized representations to smooth the domain gap by mining relations with additional augmented samples. Furthermore, a category-specific memory bank with sketch features is explored to reduce intra-class diversity in the sketch domain. Extensive experiments demonstrate that our approach notably outperforms the state-of-the-art methods in both Sketchy and TU-Berlin datasets. Our source code is publicly available at https://github.com/haowang1992/DSN.

(2 min) Non-linear (large) time warping is a challenging source of nuisance in time-series analysis. In this paper, we propose a novel diffeomorphic temporal transformer network for both pairwise and joint time-series alignment. Our ResNet-TW (Deep Residual Network for Time Warping) tackles the alignment problem by compositing a flow of incremental diffeomorphic mappings. Governed by the flow equation, our Residual Network (ResNet) builds smooth, fluid and regular flows of velocity fields and consequently generates smooth and invertible transformations (i.e. diffeomorphic warping functions). Inspired by the elegant Large Deformation Diffeomorphic Metric Mapping (LDDMM) framework, the final transformation is built by the flow of time-dependent vector fields which are none other than the building blocks of our Residual Network. The latter is naturally viewed as an Eulerian discretization schema of the flow equation (an ODE). Once trained, our ResNet-TW aligns unseen data by a single inexpensive forward pass. As we show in experiments on both univariate (84 datasets from UCR archive) and multivariate time-series (MSR Action-3D, Florence-3D and MSR Daily Activity), ResNet-TW achieves competitive performance in joint alignment and classification.

(2 min) Traditional image resizing methods usually work in pixel space and use various saliency measures. The challenge is to adjust the image shape while trying to preserve important content. In this paper we perform image resizing in feature space where the deep layers of a neural network contain rich important semantic information. We directly adjust the image feature maps, extracted from a pre-trained classification network, and reconstruct the resized image using a neural-network based optimization. This novel approach leverages the hierarchical encoding of the network, and in particular, the high-level discriminative power of its deeper layers, that recognizes semantic objects and regions and allows maintaining their aspect ratio. Our use of reconstruction from deep features diminishes the artifacts introduced by image-space resizing operators. We evaluate our method on benchmarks, compare to alternative approaches, and demonstrate its strength on challenging images.

(2 min) High quality facial image editing is a challenging problem in the movie post-production industry, requiring a high degree of control and identity preservation. Previous works that attempt to tackle this problem may suffer from the entanglement of facial attributes and the loss of the person's identity. Furthermore, many algorithms are limited to a certain task. To tackle these limitations, we propose to edit facial attributes via the latent space of a StyleGAN generator, by training a dedicated latent transformation network and incorporating explicit disentanglement and identity preservation terms in the loss function. We further introduce a pipeline to generalize our face editing to videos. Our model achieves a disentangled, controllable, and identity-preserving facial attribute editing, even in the challenging case of real (i.e., non-synthetic) images and videos. We conduct extensive experiments on image and video datasets and show that our model outperforms other state-of-the-art methods in visual quality and quantitative evaluation.

(2 min) Confidence-aware learning is proven as an effective solution to prevent networks becoming overconfident. We present a confidence-aware camouflaged object detection framework using dynamic supervision to produce both accurate camouflage map and meaningful "confidence" representing model awareness about the current prediction. A camouflaged object detection network is designed to produce our camouflage prediction. Then, we concatenate it with the input image and feed it to the confidence estimation network to produce an one channel confidence map.We generate dynamic supervision for the confidence estimation network, representing the agreement of camouflage prediction with the ground truth camouflage map. With the produced confidence map, we introduce confidence-aware learning with the confidence map as guidance to pay more attention to the hard/low-confidence pixels in the loss function. We claim that, once trained, our confidence estimation network can evaluate pixel-wise accuracy of the prediction without relying on the ground truth camouflage map. Extensive results on four camouflaged object detection testing datasets illustrate the superior performance of the proposed model in explaining the camouflage prediction.

(2 min) Geometric Deep Learning has recently made striking progress with the advent of continuous Deep Implicit Fields. They allow for detailed modeling of watertight surfaces of arbitrary topology while not relying on a 3D Euclidean grid, resulting in a learnable parameterization that is unlimited in resolution. Unfortunately, these methods are often unsuitable for applications that require an explicit mesh-based surface representation because converting an implicit field to such a representation relies on the Marching Cubes algorithm, which cannot be differentiated with respect to the underlying implicit field. In this work, we remove this limitation and introduce a differentiable way to produce explicit surface mesh representations from Deep Implicit Fields. Our key insight is that by reasoning on how implicit field perturbations impact local surface geometry, one can ultimately differentiate the 3D location of surface samples with respect to the underlying deep implicit field. We exploit this to define DeepMesh -- end-to-end differentiable mesh representation that can vary its topology. We use two different applications to validate our theoretical insight: Single view 3D Reconstruction via Differentiable Rendering and Physically-Driven Shape Optimization. In both cases our end-to-end differentiable parameterization gives us an edge over state-of-the-art algorithms.

(2 min) Automatically recognizing the layout of handwritten documents is an important step towards useful extraction of information from those documents. The most common application is to feed downstream applications such as automatic text recognition and keyword spotting; however, the recognition of the layout also helps to establish relationships between elements in the document which allows to enrich the information that can be extracted. Most of the modern document layout analysis systems are designed to address only one part of the document layout problem, namely: baseline detection or region segmentation. In contrast, we evaluate the effectiveness of the Mask-RCNN architecture to address the problem of baseline detection and region segmentation in an integrated manner. We present experimental results on two handwritten text datasets and one handwritten music dataset. The analyzed architecture yields promising results, outperforming state-of-the-art techniques in all three datasets.

(2 min) Benefited from considerable pixel-level annotations collected from a specific situation (source), the trained semantic segmentation model performs quite well, but fails in a new situation (target) due to the large domain shift. To mitigate the domain gap, previous cross-domain semantic segmentation methods always assume the co-existence of source data and target data during distribution alignment. However, the access to source data in the real scenario may raise privacy concerns and violate intellectual property. To tackle this problem, we focus on an interesting and challenging cross-domain semantic segmentation task where only the trained source model is provided to the target domain, and further propose a unified framework called Domain Adaptive Semantic Segmentation without Source data (DAS$^3$ for short). Specifically, DAS$^3$ consists of three schemes, i.e., feature alignment, self-training, and information propagation. First, we mainly develop a focal entropic loss on the network outputs to implicitly align the target features with unseen source features via the provided source model. Second, besides positive pseudo labels in vanilla self-training, we first introduce negative pseudo labels to the field and develop a bi-directional self-training strategy to enhance the representation learning in the target domain. Finally, the information propagation scheme further reduces the intra-domain discrepancy within the target domain via pseudo semi-supervised learning. Extensive results on synthesis-to-real and cross-city driving datasets validate DAS$^3$ yields state-of-the-art performance, even on par with methods that need access to source data.

(2 min) This paper presents a novel fingerprinting methodology for the Intellectual Property protection of generative models. Prior solutions for discriminative models usually adopt adversarial examples as the fingerprints, which give anomalous inference behaviors and prediction results. Hence, these methods are not stealthy and can be easily recognized by the adversary. Our approach leverages the invisible backdoor technique to overcome the above limitation. Specifically, we design verification samples, whose model outputs look normal but can trigger a backdoor classifier to make abnormal predictions. We propose a new backdoor embedding approach with Unique-Triplet Loss and fine-grained categorization to enhance the effectiveness of our fingerprints. Extensive evaluations show that this solution can outperform other strategies with higher robustness, uniqueness and stealthiness for various GAN models.

(2 min) Tracking and reconstructing the 3D pose and geometry of two hands in interaction is a challenging problem that has a high relevance for several human-computer interaction applications, including AR/VR, robotics, or sign language recognition. Existing works are either limited to simpler tracking settings (e.g., considering only a single hand or two spatially separated hands), or rely on less ubiquitous sensors, such as depth cameras. In contrast, in this work we present the first real-time method for motion capture of skeletal pose and 3D surface geometry of hands from a single RGB camera that explicitly considers close interactions. In order to address the inherent depth ambiguities in RGB data, we propose a novel multi-task CNN that regresses multiple complementary pieces of information, including segmentation, dense matchings to a 3D hand model, and 2D keypoint positions, together with newly proposed intra-hand relative depth and inter-hand distance maps. These predictions are subsequently used in a generative model fitting framework in order to estimate pose and shape parameters of a 3D hand model for both hands. We experimentally verify the individual components of our RGB two-hand tracking and 3D reconstruction pipeline through an extensive ablation study. Moreover, we demonstrate that our approach offers previously unseen two-hand tracking performance from RGB, and quantitatively and qualitatively outperforms existing RGB-based methods that were not explicitly designed for two-hand interactions. Moreover, our method even performs on-par with depth-based real-time methods.

(2 min) Multi-label image classification (MLIC) is a fundamental and practical task, which aims to assign multiple possible labels to an image. In recent years, many deep convolutional neural network (CNN) based approaches have been proposed which model label correlations to discover semantics of labels and learn semantic representations of images. This paper advances this research direction by improving both the modeling of label correlations and the learning of semantic representations. On the one hand, besides the local semantics of each label, we propose to further explore global semantics shared by multiple labels. On the other hand, existing approaches mainly learn the semantic representations at the last convolutional layer of a CNN. But it has been noted that the image representations of different layers of CNN capture different levels or scales of features and have different discriminative abilities. We thus propose to learn semantic representations at multiple convolutional layers. To this end, this paper designs a Multi-layered Semantic Representation Network (MSRN) which discovers both local and global semantics of labels through modeling label correlations and utilizes the label semantics to guide the semantic representations learning at multiple layers through an attention mechanism. Extensive experiments on four benchmark datasets including VOC 2007, COCO, NUS-WIDE, and Apparel show a competitive performance of the proposed MSRN against state-of-the-art models.

(2 min) This technical report presents our solution for temporal action detection task in AcitivityNet Challenge 2021. The purpose of this task is to locate and identify actions of interest in long untrimmed videos. The crucial challenge of the task comes from that the temporal duration of action varies dramatically, and the target actions are typically embedded in a background of irrelevant activities. Our solution builds on BMN, and mainly contains three steps: 1) action classification and feature encoding by Slowfast, CSN and ViViT; 2) proposal generation. We improve BMN by embedding the proposed Proposal Relation Network (PRN), by which we can generate proposals of high quality; 3) action detection. We calculate the detection results by assigning the proposals with corresponding classification results. Finally, we ensemble the results under different settings and achieve 44.7% on the test set, which improves the champion result in ActivityNet 2020 by 1.9% in terms of average mAP.

(2 min) UK Biobank (UKB) is conducting a large-scale study of more than half a million volunteers, collecting health-related information on genetics, lifestyle, blood biochemistry, and more. Medical imaging furthermore targets 100,000 subjects, with 70,000 follow-up sessions, enabling measurements of organs, muscle, and body composition. With up to 170,000 mounting MR images, various methodologies are accordingly engaged in large-scale image analysis. This work presents an experimental inference engine that can automatically predict a comprehensive profile of subject metadata from UKB neck-to-knee body MRI. In cross-validation, it accurately inferred baseline characteristics such as age, height, weight, and sex, but also emulated measurements of body composition by DXA, organ volumes, and abstract properties like grip strength, pulse rate, and type 2 diabetic status (AUC: 0.866). The proposed system can automatically analyze thousands of subjects within hours and provide individual confidence intervals. The underlying methodology is based on convolutional neural networks for image-based mean-variance regression on two-dimensional representations of the MRI data. This work aims to make the proposed system available for free to researchers, who can use it to obtain fast and fully-automated estimates of 72 different measurements immediately upon release of new UK Biobank image data.

(2 min) We propose a novel and effective input transformation based adversarial defense method against gray- and black-box attack, which is computationally efficient and does not require any adversarial training or retraining of a classification model. We first show that a very simple iterative Gaussian smoothing can effectively wash out adversarial noise and achieve substantially high robust accuracy. Based on the observation, we propose Self-Supervised Iterative Contextual Smoothing (SSICS), which aims to reconstruct the original discriminative features from the Gaussian-smoothed image in context-adaptive manner, while still smoothing out the adversarial noise. From the experiments on ImageNet, we show that our SSICS achieves both high standard accuracy and very competitive robust accuracy for the gray- and black-box attacks; e.g., transfer-based PGD-attack and score-based attack. A note-worthy point to stress is that our defense is free of computationally expensive adversarial training, yet, can approach its robust accuracy via input transformation.

(2 min) Skin color detection is an essential required step in various applications related to computer vision. These applications will include face detection, finding pornographic images in movies and photos, finding ethnicity, age, diagnosis, and so on. Therefore, proposing a proper skin detection method can provide solution to several problems. In this study, first a new color space is created using FCM and PSO algorithms. Then, skin classification has been performed in the new color space utilizing linear and nonlinear modes. Additionally, it has been done in RGB and LAB color spaces by using ANFIS and neural network. Skin detection in RBG color space has been performed using Mahalanobis distance and Euclidean distance algorithms. In comparison, this method has 18.38% higher accuracy than the most accurate method on the same database. Additionally, this method has achieved 90.05% in equal error rate (1-EER) in testing COMPAQ dataset and 92.93% accuracy in testing Pratheepan dataset, which compared to the previous method on COMPAQ database, 1-EER has increased by %0.87.

(2 min) We present a no-code Artificial Intelligence (AI) platform called Trinity with the main design goal of enabling both machine learning researchers and non-technical geospatial domain experts to experiment with domain-specific signals and datasets for solving a variety of complex problems on their own. This versatility to solve diverse problems is achieved by transforming complex Spatio-temporal datasets to make them consumable by standard deep learning models, in this case, Convolutional Neural Networks (CNNs), and giving the ability to formulate disparate problems in a standard way, eg. semantic segmentation. With an intuitive user interface, a feature store that hosts derivatives of complex feature engineering, a deep learning kernel, and a scalable data processing mechanism, Trinity provides a powerful platform for domain experts to share the stage with scientists and engineers in solving business-critical problems. It enables quick prototyping, rapid experimentation and reduces the time to production by standardizing model building and deployment. In this paper, we present our motivation behind Trinity and its design along with showcasing sample applications to motivate the idea of lowering the bar to using AI.

(2 min) We consider a class-incremental semantic segmentation (CISS) problem. While some recently proposed algorithms utilized variants of knowledge distillation (KD) technique to tackle the problem, they only partially addressed the key additional challenges in CISS that causes the catastrophic forgetting; i.e., the semantic drift of the background class and multi-label prediction issue. To better address these challenges, we propose a new method, dubbed as SSUL-M (Semantic Segmentation with Unknown Label with Memory), by carefully combining several techniques tailored for semantic segmentation. More specifically, we make three main contributions; (1) modeling unknown class within the background class to help learning future classes (help plasticity), (2) freezing backbone network and past classifiers with binary cross-entropy loss and pseudo-labeling to overcome catastrophic forgetting (help stability), and (3) utilizing tiny exemplar memory for the first time in CISS to improve both plasticity and stability. As a result, we show our method achieves significantly better performance than the recent state-of-the-art baselines on the standard benchmark datasets. Furthermore, we justify our contributions with thorough and extensive ablation analyses and discuss different natures of the CISS problem compared to the standard class-incremental learning for classification.

(2 min) Dysfluencies and variations in speech pronunciation can severely degrade speech recognition performance, and for many individuals with moderate-to-severe speech disorders, voice operated systems do not work. Current speech recognition systems are trained primarily with data from fluent speakers and as a consequence do not generalize well to speech with dysfluencies such as sound or word repetitions, sound prolongations, or audible blocks. The focus of this work is on quantitative analysis of a consumer speech recognition system on individuals who stutter and production-oriented approaches for improving performance for common voice assistant tasks (i.e., "what is the weather?"). At baseline, this system introduces a significant number of insertion and substitution errors resulting in intended speech Word Error Rates (isWER) that are 13.64\% worse (absolute) for individuals with fluency disorders. We show that by simply tuning the decoding parameters in an existing hybrid speech recognition system one can improve isWER by 24\% (relative) for individuals with fluency disorders. Tuning these parameters translates to 3.6\% better domain recognition and 1.7\% better intent recognition relative to the default setup for the 18 study participants across all stuttering severities.

(2 min) With the recent developments in neural networks, there has been a resurgence in algorithms for the automatic generation of simulation ready electronic circuits from hand-drawn circuits. However, most of the approaches in literature were confined to classify different types of electrical components and only a few of those methods have shown a way to rebuild the circuit schematic from the scanned image, which is extremely important for further automation of netlist generation. This paper proposes a real-time algorithm for the automatic recognition of hand-drawn electrical circuits based on object detection and circuit node recognition. The proposed approach employs You Only Look Once version 5 (YOLOv5) for detection of circuit components and a novel Hough transform based approach for node recognition. Using YOLOv5 object detection algorithm, a mean average precision (mAP0.5) of 98.2% is achieved in detecting the components. The proposed method is also able to rebuild the circuit schematic with 80% accuracy.

(2 min) In many computer vision classification tasks, class priors at test time often differ from priors on the training set. In the case of such prior shift, classifiers must be adapted correspondingly to maintain close to optimal performance. This paper analyzes methods for adaptation of probabilistic classifiers to new priors and for estimating new priors on an unlabeled test set. We propose a novel method to address a known issue of prior estimation methods based on confusion matrices, where inconsistent estimates of decision probabilities and confusion matrices lead to negative values in the estimated priors. Experiments on fine-grained image classification datasets provide insight into the best practice of prior shift estimation and classifier adaptation and show that the proposed method achieves state-of-the-art results in prior adaptation. Applying the best practice to two tasks with naturally imbalanced priors, learning from web-crawled images and plant species classification, increased the recognition accuracy by 1.1% and 3.4% respectively.

(2 min) Intelligent systems are increasingly part of our everyday lives and have been integrated seamlessly to the point where it is difficult to imagine a world without them. Physical manifestations of those systems on the other hand, in the form of embodied agents or robots, have so far been used only for specific applications and are often limited to functional roles (e.g. in the industry, entertainment and military fields). Given the current growth and innovation in the research communities concerned with the topics of robot navigation, human-robot-interaction and human activity recognition, it seems like this might soon change. Robots are increasingly easy to obtain and use and the acceptance of them in general is growing. However, the design of a socially compliant robot that can function as a companion needs to take various areas of research into account. This paper is concerned with the navigation aspect of a socially-compliant robot and provides a survey of existing solutions for the relevant areas of research as well as an outlook on possible future directions.

(2 min) Light field technology has increasingly attracted the attention of the research community with its many possible applications. The lenslet array in commercial plenoptic cameras helps capture both the spatial and angular information of light rays in a single exposure. While the resulting high dimensionality of light field data enables its superior capabilities, it also impedes its extensive adoption. Hence, there is a compelling need for efficient compression of light field images. Existing solutions are commonly composed of several separate modules, some of which may not have been designed for the specific structure and quality of light field data. This increases the complexity of the codec and results in impractical decoding runtimes. We propose a new learning-based, disparity-aided model for compression of 4D light field images capable of parallel decoding. The model is end-to-end trainable, eliminating the need for hand-tuning separate modules and allowing joint learning of rate and distortion. The disparity-aided approach ensures the structural integrity of the reconstructed light fields. Comparisons with the state of the art show encouraging performance in terms of PSNR and MS-SSIM metrics. Also, there is a notable gain in the encoding and decoding runtimes. Source code is available at https://moha23.github.io/LFDAAE.

(2 min) We address the problem of universal domain adaptation (UDA) in ordinal regression (OR), which attempts to solve classification problems in which labels are not independent, but follow a natural order. We show that the UDA techniques developed for classification and based on the clustering assumption, under-perform in OR settings. We propose a method that complements the OR classifier with an auxiliary task of order learning, which plays the double role of discriminating between common and private instances, and expanding class labels to the private target images via ranking. Combined with adversarial domain discrimination, our model is able to address the closed set, partial and open set configurations. We evaluate our method on three face age estimation datasets, and show that it outperforms the baseline methods.

(2 min) Generic Event Boundary Detection (GEBD) is a newly introduced task that aims to detect "general" event boundaries that correspond to natural human perception. In this paper, we introduce a novel contrastive learning based approach to deal with the GEBD. Our intuition is that the feature similarity of the video snippet would significantly vary near the event boundaries, while remaining relatively the same in the remaining part of the video. In our model, Temporal Self-similarity Matrix (TSM) is utilized as an intermediate representation which takes on a role as an information bottleneck. With our model, we achieved significant performance boost compared to the given baselines. Our code is available at https://github.com/hello-jinwoo/LOVEU-CVPR2021.

(2 min) With leveraging the weight-sharing and continuous relaxation to enable gradient-descent to alternately optimize the supernet weights and the architecture parameters through a bi-level optimization paradigm, \textit{Differentiable ARchiTecture Search} (DARTS) has become the mainstream method in Neural Architecture Search (NAS) due to its simplicity and efficiency. However, more recent works found that the performance of the searched architecture barely increases with the optimization proceeding in DARTS. In addition, several concurrent works show that the NAS could find more competitive architectures without labels. The above observations reveal that the supervision signal in DARTS may be a poor indicator for architecture optimization, inspiring a foundational question: instead of using the supervision signal to perform bi-level optimization, \textit{can we find high-quality architectures \textbf{without any training nor labels}}? We provide an affirmative answer by customizing the NAS as a network pruning at initialization problem. By leveraging recent techniques on the network pruning at initialization, we designed a FreeFlow proxy to score the importance of candidate operations in NAS without any training nor labels, and proposed a novel framework called \textit{training and label free neural architecture search} (\textbf{FreeNAS}) accordingly. We show that, without any training nor labels, FreeNAS with the proposed FreeFlow proxy can outperform most NAS baselines. More importantly, our framework is extremely efficient, which completes the architecture search within only \textbf{3.6s} and \textbf{79s} on a single GPU for the NAS-Bench-201 and DARTS search space, respectively. We hope our work inspires more attempts in solving NAS from the perspective of pruning at initialization.

(2 min) LiDAR-based SLAM system is admittedly more accurate and stable than others, while its loop closure detection is still an open issue. With the development of 3D semantic segmentation for point cloud, semantic information can be obtained conveniently and steadily, essential for high-level intelligence and conductive to SLAM. In this paper, we present a novel semantic-aided LiDAR SLAM with loop closure based on LOAM, named SA-LOAM, which leverages semantics in odometry as well as loop closure detection. Specifically, we propose a semantic-assisted ICP, including semantically matching, downsampling and plane constraint, and integrates a semantic graph-based place recognition method in our loop closure detection module. Benefitting from semantics, we can improve the localization accuracy, detect loop closures effectively, and construct a global consistent semantic map even in large-scale scenes. Extensive experiments on KITTI and Ford Campus dataset show that our system significantly improves baseline performance, has generalization ability to unseen data and achieves competitive results compared with state-of-the-art methods.

(2 min) We present DocFormer -- a multi-modal transformer based architecture for the task of Visual Document Understanding (VDU). VDU is a challenging problem which aims to understand documents in their varied formats (forms, receipts etc.) and layouts. In addition, DocFormer is pre-trained in an unsupervised fashion using carefully designed tasks which encourage multi-modal interaction. DocFormer uses text, vision and spatial features and combines them using a novel multi-modal self-attention layer. DocFormer also shares learned spatial embeddings across modalities which makes it easy for the model to correlate text to visual tokens and vice versa. DocFormer is evaluated on 4 different datasets each with strong baselines. DocFormer achieves state-of-the-art results on all of them, sometimes beating models 4x its size (in no. of parameters).

(2 min) We introduce an approach that accurately reconstructs 3D human poses and detailed 3D full-body geometric models from single images in realtime. The key idea of our approach is a novel end-to-end multi-task deep learning framework that uses single images to predict five outputs simultaneously: foreground segmentation mask, 2D joints positions, semantic body partitions, 3D part orientations and uv coordinates (uv map). The multi-task network architecture not only generates more visual cues for reconstruction, but also makes each individual prediction more accurate. The CNN regressor is further combined with an optimization based algorithm for accurate kinematic pose reconstruction and full-body shape modeling. We show that the realtime reconstruction reaches accurate fitting that has not been seen before, especially for wild images. We demonstrate the results of our realtime 3D pose and human body reconstruction system on various challenging in-the-wild videos. We show the system advances the frontier of 3D human body and pose reconstruction from single images by quantitative evaluations and comparisons with state-of-the-art methods.

(2 min) Recent advances in bioimaging have provided scientists a superior high spatial-temporal resolution to observe dynamics of living cells as 3D volumetric videos. Unfortunately, the 3D biomedical video analysis is lagging, impeded by resource insensitive human curation using off-the-shelf 3D analytic tools. Herein, biologists often need to discard a considerable amount of rich 3D spatial information by compromising on 2D analysis via maximum intensity projection. Recently, pixel embedding-based cell instance segmentation and tracking provided a neat and generalizable computing paradigm for understanding cellular dynamics. In this work, we propose a novel spatial-temporal voxel-embedding (VoxelEmbed) based learning method to perform simultaneous cell instance segmenting and tracking on 3D volumetric video sequences. Our contribution is in four-fold: (1) The proposed voxel embedding generalizes the pixel embedding with 3D context information; (2) Present a simple multi-stream learning approach that allows effective spatial-temporal embedding; (3) Accomplished an end-to-end framework for one-stage 3D cell instance segmentation and tracking without heavy parameter tuning; (4) The proposed 3D quantification is memory efficient via a single GPU with 12 GB memory. We evaluate our VoxelEmbed method on four 3D datasets (with different cell types) from the ISBI Cell Tracking Challenge. The proposed VoxelEmbed method achieved consistent superior overall performance (OP) on two densely annotated datasets. The performance is also competitive on two sparsely annotated cohorts with 20.6% and 2% of data-set having segmentation annotations. The results demonstrate that the VoxelEmbed method is a generalizable and memory-efficient solution.

(2 min) In Argoverse motion forecasting competition, the task is to predict the probabilistic future trajectory distribution for the interested targets in the traffic scene. We use vectorized lane map and 2 s targets' history trajectories as input. Then the model outputs 6 forecasted trajectories with probability for each target.

(2 min) Textures contain a wealth of image information and are widely used in various fields such as computer graphics and computer vision. With the development of machine learning, the texture synthesis and generation have been greatly improved. As a very common element in everyday life, wallpapers contain a wealth of texture information, making it difficult to annotate with a simple single label. Moreover, wallpaper designers spend significant time to create different styles of wallpaper. For this purpose, this paper proposes to describe wallpaper texture images by using multi-label semantics. Based on these labels and generative adversarial networks, we present a framework for perception driven wallpaper texture generation and style transfer. In this framework, a perceptual model is trained to recognize whether the wallpapers produced by the generator network are sufficiently realistic and have the attribute designated by given perceptual description; these multi-label semantic attributes are treated as condition variables to generate wallpaper images. The generated wallpaper images can be converted to those with well-known artist styles using CycleGAN. Finally, using the aesthetic evaluation method, the generated wallpaper images are quantitatively measured. The experimental results demonstrate that the proposed method can generate wallpaper textures conforming to human aesthetics and have artistic characteristics.

(2 min) Kernel segmentation aims at partitioning a data sequence into several non-overlapping segments that may have nonlinear and complex structures. In general, it is formulated as a discrete optimization problem with combinatorial constraints. A popular algorithm for optimally solving this problem is dynamic programming (DP), which has quadratic computation and memory requirements. Given that sequences in practice are too long, this algorithm is not a practical approach. Although many heuristic algorithms have been proposed to approximate the optimal segmentation, they have no guarantee on the quality of their solutions. In this paper, we take a differentiable approach to alleviate the aforementioned issues. First, we introduce a novel sigmoid-based regularization to smoothly approximate the combinatorial constraints. Combining it with objective of the balanced kernel clustering, we formulate a differentiable model termed Kernel clustering with sigmoid-based regularization (KCSR), where the gradient-based algorithm can be exploited to obtain the optimal segmentation. Second, we develop a stochastic variant of the proposed model. By using the stochastic gradient descent algorithm, which has much lower time and space complexities, for optimization, the second model can perform segmentation on overlong data sequences. Finally, for simultaneously segmenting multiple data sequences, we slightly modify the sigmoid-based regularization to further introduce an extended variant of the proposed model. Through extensive experiments on various types of data sequences performances of our models are evaluated and compared with those of the existing methods. The experimental results validate advantages of the proposed models. Our Matlab source code is available on github.

(2 min) Place Recognition is a crucial capability for mobile robot localization and navigation. Image-based or Visual Place Recognition (VPR) is a challenging problem as scene appearance and camera viewpoint can change significantly when places are revisited. Recent VPR methods based on ``sequential representations'' have shown promising results as compared to traditional sequence score aggregation or single image based techniques. In parallel to these endeavors, 3D point clouds based place recognition is also being explored following the advances in deep learning based point cloud processing. However, a key question remains: is an explicit 3D structure based place representation always superior to an implicit ``spatial'' representation based on sequence of RGB images which can inherently learn scene structure. In this extended abstract, we attempt to compare these two types of methods by considering a similar ``metric span'' to represent places. We compare a 3D point cloud based method (PointNetVLAD) with image sequence based methods (SeqNet and others) and showcase that image sequence based techniques approach, and can even surpass, the performance achieved by point cloud based methods for a given metric span. These performance variations can be attributed to differences in data richness of input sensors as well as data accumulation strategies for a mobile robot. While a perfect apple-to-apple comparison may not be feasible for these two different modalities, the presented comparison takes a step in the direction of answering deeper questions regarding spatial representations, relevant to several applications like Autonomous Driving and Augmented/Virtual Reality. Source code available publicly https://github.com/oravus/seqNet.

(2 min) Image reconstruction is likely the most predominant auxiliary task for image classification. In this paper, we investigate ``estimating the Fourier Transform of the input image" as a potential alternative auxiliary task, in the hope that it may further boost the performances on the primary task or introduce novel constraints not well covered by image reconstruction. We experimented with five popular classification architectures on the CIFAR-10 dataset, and the empirical results indicated that our proposed auxiliary task generally improves the classification accuracy. More notably, the results showed that in certain cases our proposed auxiliary task may enhance the classifiers' resistance to adversarial attacks generated using the fast gradient sign method.

(2 min) High-resolution satellite imagery has proven useful for a broad range of tasks, including measurement of global human population, local economic livelihoods, and biodiversity, among many others. Unfortunately, high-resolution imagery is both infrequently collected and expensive to purchase, making it hard to efficiently and effectively scale these downstream tasks over both time and space. We propose a new conditional pixel synthesis model that uses abundant, low-cost, low-resolution imagery to generate accurate high-resolution imagery at locations and times in which it is unavailable. We show that our model attains photo-realistic sample quality and outperforms competing baselines on a key downstream task -- object counting -- particularly in geographic locations where conditions on the ground are changing rapidly.

(2 min) Most modern neural networks for classification fail to take into account the concept of the unknown. Trained neural networks are usually tested in an unrealistic scenario with only examples from a closed set of known classes. In an attempt to develop a more realistic model, the concept of working in an open set environment has been introduced. This in turn leads to the concept of incremental learning where a model with its own architecture and initial trained set of data can identify unknown classes during the testing phase and autonomously update itself if evidence of a new class is detected. Some problems that arise in incremental learning are inefficient use of resources to retrain the classifier repeatedly and the decrease of classification accuracy as multiple classes are added over time. This process of instantiating new classes is repeated as many times as necessary, accruing errors. To address these problems, this paper proposes the Classification Confidence Threshold approach to prime neural networks for incremental learning to keep accuracies high by limiting forgetting. A lean method is also used to reduce resources used in the retraining of the neural network. The proposed method is based on the idea that a network is able to incrementally learn a new class even when exposed to a limited number samples associated with the new class. This method can be applied to most existing neural networks with minimal changes to network architecture.

(2 min) Image Processing algorithms for vision-based navigation require reliable image simulation capacities. In this paper we explain why traditional rendering engines may present limitations that are potentially critical for space applications. We introduce Airbus SurRender software v7 and provide details on features that make it a very powerful space image simulator. We show how SurRender is at the heart of the development processes of our computer vision solutions and we provide a series of illustrations of rendered images for various use cases ranging from Moon and Solar System exploration, to in orbit rendezvous and planetary robotics.

(2 min) Bilevel optimization recently has attracted increased interest in machine learning due to its many applications such as hyper-parameter optimization and policy optimization. Although some methods recently have been proposed to solve the bilevel problems, these methods do not consider using adaptive learning rates. To fill this gap, in the paper, we propose a class of fast and effective adaptive methods for solving bilevel optimization problems that the outer problem is possibly nonconvex and the inner problem is strongly-convex. Specifically, we propose a fast single-loop BiAdam algorithm based on the basic momentum technique, which achieves a sample complexity of $\tilde{O}(\epsilon^{-4})$ for finding an $\epsilon$-stationary point. At the same time, we propose an accelerated version of BiAdam algorithm (VR-BiAdam) by using variance reduced technique, which reaches the best known sample complexity of $\tilde{O}(\epsilon^{-3})$. To further reduce computation in estimating derivatives, we propose a fast single-loop stochastic approximated BiAdam algorithm (saBiAdam) by avoiding the Hessian inverse, which still achieves a sample complexity of $\tilde{O}(\epsilon^{-4})$ without large batches. We further present an accelerated version of saBiAdam algorithm (VR-saBiAdam), which also reaches the best known sample complexity of $\tilde{O}(\epsilon^{-3})$. We apply the unified adaptive matrices to our methods as the SUPER-ADAM \citep{huang2021super}, which including many types of adaptive learning rates. Moreover, our framework can flexibly use the momentum and variance reduced techniques. In particular, we provide a useful convergence analysis framework for both the constrained and unconstrained bilevel optimization. To the best of our knowledge, we first study the adaptive bilevel optimization methods with adaptive learning rates.

(2 min) The author of this work proposes an overview of the recent semi-supervised learning approaches and related works. Despite the remarkable success of neural networks in various applications, there exist few formidable constraints including the need for a large amount of labeled data. Therefore, semi-supervised learning, which is a learning scheme in which the scarce labels and a larger amount of unlabeled data are utilized to train models (e.g., deep neural networks) is getting more important. Based on the key assumptions of semi-supervised learning, which are the manifold assumption, cluster assumption, and continuity assumption, the work reviews the recent semi-supervised learning approaches. In particular, the methods in regard to using deep neural networks in a semi-supervised learning setting are primarily discussed. In addition, the existing works are first classified based on the underlying idea and explained, and then the holistic approaches that unify the aforementioned ideas are detailed.

(2 min) We propose unsupervised embedding adaptation for the downstream few-shot classification task. Based on findings that deep neural networks learn to generalize before memorizing, we develop Early-Stage Feature Reconstruction (ESFR) -- a novel adaptation scheme with feature reconstruction and dimensionality-driven early stopping that finds generalizable features. Incorporating ESFR consistently improves the performance of baseline methods on all standard settings, including the recently proposed transductive method. ESFR used in conjunction with the transductive method further achieves state-of-the-art performance on mini-ImageNet, tiered-ImageNet, and CUB; especially with 1.2%~2.0% improvements in accuracy over the previous best performing method on 1-shot setting.

(2 min) Unsupervised domain adaptation is used in many machine learning applications where, during training, a model has access to unlabeled data in the target domain, and a related labeled dataset. In this paper, we introduce a novel and general domain-adversarial framework. Specifically, we derive a novel generalization bound for domain adaptation that exploits a new measure of discrepancy between distributions based on a variational characterization of f-divergences. It recovers the theoretical results from Ben-David et al. (2010a) as a special case and supports divergences used in practice. Based on this bound, we derive a new algorithmic framework that introduces a key correction in the original adversarial training method of Ganin et al. (2016). We show that many regularizers and ad-hoc objectives introduced over the last years in this framework are then not required to achieve performance comparable to (if not better than) state-of-the-art domain-adversarial methods. Experimental analysis conducted on real-world natural language and computer vision datasets show that our framework outperforms existing baselines, and obtains the best results for f-divergences that were not considered previously in domain-adversarial learning.

(2 min) The advent of social media platforms has been a catalyst for the development of digital photography that engendered a boom in vision applications. With this motivation, we introduce a large-scale dataset termed 'Photozilla', which includes over 990k images belonging to 10 different photographic styles. The dataset is then used to train 3 classification models to automatically classify the images into the relevant style which resulted in an accuracy of ~96%. With the rapid evolution of digital photography, we have seen new types of photography styles emerging at an exponential rate. On that account, we present a novel Siamese-based network that uses the trained classification models as the base architecture to adapt and classify unseen styles with only 25 training samples. We report an accuracy of over 68% for identifying 10 other distinct types of photography styles. This dataset can be found at https://trisha025.github.io/Photozilla/

(2 min) While working with fingerprint images acquired from crime scenes, mobile cameras, or low-quality sensors, it becomes difficult for automated identification systems to verify the identity due to image blur and distortion. We propose a fingerprint deblurring model FDeblur-GAN, based on the conditional Generative Adversarial Networks (cGANs) and multi-stage framework of the stack GAN. Additionally, we integrate two auxiliary sub-networks into the model for the deblurring task. The first sub-network is a ridge extractor model. It is added to generate ridge maps to ensure that fingerprint information and minutiae are preserved in the deblurring process and prevent the model from generating erroneous minutiae. The second sub-network is a verifier that helps the generator to preserve the ID information during the generation process. Using a database of blurred fingerprints and corresponding ridge maps, the deep network learns to deblur from the input blurry samples. We evaluate the proposed method in combination with two different fingerprint matching algorithms. We achieved an accuracy of 95.18% on our fingerprint database for the task of matching deblurred and ground truth fingerprints.

(2 min) Accurate liver and lesion segmentation from computed tomography (CT) images are highly demanded in clinical practice for assisting the diagnosis and assessment of hepatic tumor disease. However, automatic liver and lesion segmentation from contrast-enhanced CT volumes is extremely challenging due to the diversity in contrast, resolution, and quality of images. Previous methods based on UNet for 2D slice-by-slice or 3D volume-by-volume segmentation either lack sufficient spatial contexts or suffer from high GPU computational cost, which limits the performance. To tackle these issues, we propose a novel context-aware PolyUNet for accurate liver and lesion segmentation. It jointly explores structural diversity and consecutive t-adjacent slices to enrich feature expressive power and spatial contextual information while avoiding the overload of GPU memory consumption. In addition, we utilize zoom out/in and two-stage refinement strategy to exclude the irrelevant contexts and focus on the specific region for the fine-grained segmentation. Our method achieved very competitive performance at the MICCAI 2017 Liver Tumor Segmentation (LiTS) Challenge among all tasks with a single model and ranked the $3^{rd}$, $12^{th}$, $2^{nd}$, and $5^{th}$ places in the liver segmentation, lesion segmentation, lesion detection, and tumor burden estimation, respectively.

(2 min) The UN-Habitat estimates that over one billion people live in slums around the world. However, state-of-the-art techniques to detect the location of slum areas employ high-resolution satellite imagery, which is costly to obtain and process. As a result, researchers have started to look at utilising free and open-access medium resolution satellite imagery. Yet, there is no clear consensus on which data preparation and machine learning approaches are the most appropriate to use with such imagery data. In this paper, we evaluate two techniques (multi-spectral data and grey-level co-occurrence matrix feature extraction) on an open-access dataset consisting of labelled Sentinel-2 images with a spatial resolution of 10 meters. Both techniques were paired with a canonical correlation forests classifier. The results show that the grey-level co-occurrence matrix performed better than multi-spectral data for all four cities. It had an average accuracy for the slum class of 97% and a mean intersection over union of 94%, while multi-spectral data had 75% and 64% for the respective metrics. These results indicate that open-access satellite imagery with a resolution of at least 10 meters may be suitable for keeping track of development goals such as the detection of slums in cities.

(2 min) We introduce a highly robust GAN-based framework for digitizing a normalized 3D avatar of a person from a single unconstrained photo. While the input image can be of a smiling person or taken in extreme lighting conditions, our method can reliably produce a high-quality textured model of a person's face in neutral expression and skin textures under diffuse lighting condition. Cutting-edge 3D face reconstruction methods use non-linear morphable face models combined with GAN-based decoders to capture the likeness and details of a person but fail to produce neutral head models with unshaded albedo textures which is critical for creating relightable and animation-friendly avatars for integration in virtual environments. The key challenges for existing methods to work is the lack of training and ground truth data containing normalized 3D faces. We propose a two-stage approach to address this problem. First, we adopt a highly robust normalized 3D face generator by embedding a non-linear morphable face model into a StyleGAN2 network. This allows us to generate detailed but normalized facial assets. This inference is then followed by a perceptual refinement step that uses the generated assets as regularization to cope with the limited available training samples of normalized faces. We further introduce a Normalized Face Dataset, which consists of a combination photogrammetry scans, carefully selected photographs, and generated fake people with neutral expressions in diffuse lighting conditions. While our prepared dataset contains two orders of magnitude less subjects than cutting edge GAN-based 3D facial reconstruction methods, we show that it is possible to produce high-quality normalized face models for very challenging unconstrained input images, and demonstrate superior performance to the current state-of-the-art.

(2 min) Moving Object Detection (MOD) is a crucial task for the Autonomous Driving pipeline. MOD is usually handled via 2-stream convolutional architectures that incorporates both appearance and motion cues, without considering the inter-relations between the spatial or motion features. In this paper, we tackle this problem through multi-head attention mechanisms, both across the spatial and motion streams. We propose MODETR; a Moving Object DEtection TRansformer network, comprised of multi-stream transformer encoders for both spatial and motion modalities, and an object transformer decoder that produces the moving objects bounding boxes using set predictions. The whole architecture is trained end-to-end using bi-partite loss. Several methods of incorporating motion cues with the Transformer model are explored, including two-stream RGB and Optical Flow (OF) methods, and multi-stream architectures that take advantage of sequence information. To incorporate the temporal information, we propose a new Temporal Positional Encoding (TPE) approach to extend the Spatial Positional Encoding(SPE) in DETR. We explore two architectural choices for that, balancing between speed and time. To evaluate the our network, we perform the MOD task on the KITTI MOD [6] data set. Results show significant 5% mAP of the Transformer network for MOD over the state-of-the art methods. Moreover, the proposed TPE encoding provides 10% mAP improvement over the SPE baseline.

(2 min) Transfer learning with pre-training on large-scale datasets has played an increasingly significant role in computer vision and natural language processing recently. However, as there exist numerous application scenarios that have distinctive demands such as certain latency constraints and specialized data distributions, it is prohibitively expensive to take advantage of large-scale pre-training for per-task requirements. In this paper, we focus on the area of object detection and present a transfer learning system named GAIA, which could automatically and efficiently give birth to customized solutions according to heterogeneous downstream needs. GAIA is capable of providing powerful pre-trained weights, selecting models that conform to downstream demands such as latency constraints and specified data domains, and collecting relevant data for practitioners who have very few datapoints for their tasks. With GAIA, we achieve promising results on COCO, Objects365, Open Images, Caltech, CityPersons, and UODB which is a collection of datasets including KITTI, VOC, WiderFace, DOTA, Clipart, Comic, and more. Taking COCO as an example, GAIA is able to efficiently produce models covering a wide range of latency from 16ms to 53ms, and yields AP from 38.2 to 46.5 without whistles and bells. To benefit every practitioner in the community of object detection, GAIA is released at https://github.com/GAIA-vision.

(2 min) This open-source book represents our attempt to make deep learning approachable, teaching readers the concepts, the context, and the code. The entire book is drafted in Jupyter notebooks, seamlessly integrating exposition figures, math, and interactive examples with self-contained code. Our goal is to offer a resource that could (i) be freely available for everyone; (ii) offer sufficient technical depth to provide a starting point on the path to actually becoming an applied machine learning scientist; (iii) include runnable code, showing readers how to solve problems in practice; (iv) allow for rapid updates, both by us and also by the community at large; (v) be complemented by a forum for interactive discussion of technical details and to answer questions.

(2 min) Top-down attention allows neural networks, both artificial and biological, to focus on the information most relevant for a given task. This is known to enhance performance in visual perception. But it remains unclear how attention brings about its perceptual boost, especially when it comes to naturalistic settings like recognising an object in an everyday scene. What aspects of a visual task does attention help to deal with? We aim to answer this with a computational experiment based on a general framework called task-oriented ablation design. First we define a broad range of visual tasks and identify six factors that underlie task variability. Then on each task we compare the performance of two neural networks, one with top-down attention and one without. These comparisons reveal the task-dependence of attention's perceptual boost, giving a clearer idea of the role attention plays. Whereas many existing cognitive accounts link attention to stimulus-level variables, such as visual clutter and object scale, we find greater explanatory power in system-level variables that capture the interaction between the model, the distribution of training data and the task format. This finding suggests a shift in how attention is studied could be fruitful. We make publicly available our code and results, along with statistics relevant to ImageNet-based experiments beyond this one. Our contribution serves to support the development of more human-like vision models and the design of more informative machine-learning experiments.

(2 min) With the rise of automation, unmanned vehicles became a hot topic both as commercial products and as a scientific research topic. It composes a multi-disciplinary field of robotics that encompasses embedded systems, control theory, path planning, Simultaneous Localization and Mapping (SLAM), scene reconstruction, and pattern recognition. In this work, we present our exploratory research of how sensor data fusion and state-of-the-art machine learning algorithms can perform the Embodied Artificial Intelligence (E-AI) task called Visual Semantic Navigation. This task, a.k.a Object-Goal Navigation (ObjectNav) consists of autonomous navigation using egocentric visual observations to reach an object belonging to the target semantic class without prior knowledge of the environment. Our method reached fourth place on the Habitat Challenge 2021 ObjectNav on the Minival phase and the Test-Standard Phase.

(2 min) Coronary X-ray angiography is a crucial clinical procedure for the diagnosis and treatment of coronary artery disease, which accounts for roughly 16% of global deaths every year. However, the images acquired in these procedures have low resolution and poor contrast, making lesion detection and assessment challenging. Accurate coronary artery segmentation not only helps mitigate these problems, but also allows the extraction of relevant anatomical features for further analysis by quantitative methods. Although automated segmentation of coronary arteries has been proposed before, previous approaches have used non-optimal segmentation criteria, leading to less useful results. Most methods either segment only the major vessel, discarding important information from the remaining ones, or segment the whole coronary tree based mostly on contrast information, producing a noisy output that includes vessels that are not relevant for diagnosis. We adopt a better-suited clinical criterion and segment vessels according to their clinical relevance. Additionally, we simultaneously perform catheter segmentation, which may be useful for diagnosis due to the scale factor provided by the catheter's known diameter, and is a task that has not yet been performed with good results. To derive the optimal approach, we conducted an extensive comparative study of encoder-decoder architectures trained on a combination of focal loss and a variant of generalized dice loss. Based on the EfficientNet and the UNet++ architectures, we propose a line of efficient and high-performance segmentation models using a new decoder architecture, the EfficientUNet++, whose best-performing version achieved average dice scores of 0.8904 and 0.7526 for the artery and catheter classes, respectively, and an average generalized dice score of 0.9234.

(2 min) Gait analysis is an important aspect of clinical investigation for detecting neurological and musculoskeletal disorders and assessing the global health of a patient. In this paper we propose to focus our attention on extracting relevant curvature information from the body surface provided by a depth camera. We assumed that the 3D mesh was made available in a previous step and demonstrated how curvature maps could be useful to assess asymmetric anomalies with two simple simulated abnormal gaits compared with a normal one. This research set the grounds for the future development of a curvature-based gait analysis system for healthcare professionals.

(2 min) Moving objects have special importance for Autonomous Driving tasks. Detecting moving objects can be posed as Moving Object Segmentation, by segmenting the object pixels, or Moving Object Detection, by generating a bounding box for the moving targets. In this paper, we present a Multi-Task Learning architecture, based on Transformers, to jointly perform both tasks through one network. Due to the importance of the motion features to the task, the whole setup is based on a Spatio-Temporal aggregation. We evaluate the performance of the individual tasks architecture versus the MTL setup, both with early shared encoders, and late shared encoder-decoder transformers. For the latter, we present a novel joint tasks query decoder transformer, that enables us to have tasks dedicated heads out of the shared model. To evaluate our approach, we use the KITTI MOD [29] data set. Results show1.5% mAP improvement for Moving Object Detection, and 2%IoU improvement for Moving Object Segmentation, over the individual tasks networks.

(2 min) Due to high annotation costs making the best use of existing human-created training data is an important research direction. We, therefore, carry out a systematic evaluation of transferability of BERT-based neural ranking models across five English datasets. Previous studies focused primarily on zero-shot and few-shot transfer from a large dataset to a dataset with a small number of queries. In contrast, each of our collections has a substantial number of queries, which enables a full-shot evaluation mode and improves reliability of our results. Furthermore, since source datasets licences often prohibit commercial use, we compare transfer learning to training on pseudo-labels generated by a BM25 scorer. We find that training on pseudo-labels -- possibly with subsequent fine-tuning using a modest number of annotated queries -- can produce a competitive or better model compared to transfer learning. Yet, it is necessary to improve the stability and/or effectiveness of the few-shot training, which, sometimes, can degrade performance of a pretrained model.

(2 min) The Turing Award is recognized as the most influential and prestigious award in the field of computer science(CS). With the rise of the science of science (SciSci), a large amount of bibliographic data has been analyzed in an attempt to understand the hidden mechanism of scientific evolution. These include the analysis of the Nobel Prize, including physics, chemistry, medicine, etc. In this article, we extract and analyze the data of 72 Turing Award laureates from the complete bibliographic data, fill the gap in the lack of Turing Award analysis, and discover the development characteristics of computer science as an independent discipline. First, we show most Turing Award laureates have long-term and high-quality educational backgrounds, and more than 61% of them have a degree in mathematics, which indicates that mathematics has played a significant role in the development of computer science. Secondly, the data shows that not all scholars have high productivity and high h-index; that is, the number of publications and h-index is not the leading indicator for evaluating the Turing Award. Third, the average age of awardees has increased from 40 to around 70 in recent years. This may be because new breakthroughs take longer, and some new technologies need time to prove their influence. Besides, we have also found that in the past ten years, international collaboration has experienced explosive growth, showing a new paradigm in the form of collaboration. It is also worth noting that in recent years, the emergence of female winners has also been eye-catching. Finally, by analyzing the personal publication records, we find that many people are more likely to publish high-impact articles during their high-yield periods.

(2 min) In this work, we train the first monolingual Lithuanian transformer model on a relatively large corpus of Lithuanian news articles and compare various output decoding algorithms for abstractive news summarization. We achieve an average ROUGE-2 score 0.163, generated summaries are coherent and look impressive at first glance. However, some of them contain misleading information that is not so easy to spot. We describe all the technical details and share our trained model and accompanying code in an online open-source repository, as well as some characteristic samples of the generated summaries.

(2 min) Topic modeling is an unsupervised method for revealing the hidden semantic structure of a corpus. It has been increasingly widely adopted as a tool in the social sciences, including political science, digital humanities and sociological research in general. One desirable property of topic models is to allow users to find topics describing a specific aspect of the corpus. A possible solution is to incorporate domain-specific knowledge into topic modeling, but this requires a specification from domain experts. We propose a novel query-driven topic model that allows users to specify a simple query in words or phrases and return query-related topics, thus avoiding tedious work from domain experts. Our proposed approach is particularly attractive when the user-specified query has a low occurrence in a text corpus, making it difficult for traditional topic models built on word cooccurrence patterns to identify relevant topics. Experimental results demonstrate the effectiveness of our model in comparison with both classical topic models and neural topic models.

(2 min) Place Recognition is a crucial capability for mobile robot localization and navigation. Image-based or Visual Place Recognition (VPR) is a challenging problem as scene appearance and camera viewpoint can change significantly when places are revisited. Recent VPR methods based on ``sequential representations'' have shown promising results as compared to traditional sequence score aggregation or single image based techniques. In parallel to these endeavors, 3D point clouds based place recognition is also being explored following the advances in deep learning based point cloud processing. However, a key question remains: is an explicit 3D structure based place representation always superior to an implicit ``spatial'' representation based on sequence of RGB images which can inherently learn scene structure. In this extended abstract, we attempt to compare these two types of methods by considering a similar ``metric span'' to represent places. We compare a 3D point cloud based method (PointNetVLAD) with image sequence based methods (SeqNet and others) and showcase that image sequence based techniques approach, and can even surpass, the performance achieved by point cloud based methods for a given metric span. These performance variations can be attributed to differences in data richness of input sensors as well as data accumulation strategies for a mobile robot. While a perfect apple-to-apple comparison may not be feasible for these two different modalities, the presented comparison takes a step in the direction of answering deeper questions regarding spatial representations, relevant to several applications like Autonomous Driving and Augmented/Virtual Reality. Source code available publicly https://github.com/oravus/seqNet.

(2 min) Mathematical notation, i.e., the writing system used to communicate concepts in mathematics, encodes valuable information for a variety of information search and retrieval systems. Yet, mathematical notations remain mostly unutilized by today's systems. In this paper, we present the first in-depth study on the distributions of mathematical notation in two large scientific corpora: the open access arXiv (2.5B mathematical objects) and the mathematical reviewing service for pure and applied mathematics zbMATH (61M mathematical objects). Our study lays a foundation for future research projects on mathematical information retrieval for large scientific corpora. Further, we demonstrate the relevance of our results to a variety of use-cases. For example, to assist semantic extraction systems, to improve scientific search engines, and to facilitate specialized math recommendation systems. The contributions of our presented research are as follows: (1) we present the first distributional analysis of mathematical formulae on arXiv and zbMATH; (2) we retrieve relevant mathematical objects for given textual search queries (e.g., linking $P_{n}^{(\alpha, \beta)}\!\left(x\right)$ with `Jacobi polynomial'); (3) we extend zbMATH's search engine by providing relevant mathematical formulae; and (4) we exemplify the applicability of the results by presenting auto-completion for math inputs as the first contribution to math recommendation systems. To expedite future research projects, we have made available our source code and data.

(2 min) Natural Language Processing (NLP) and Information Retrieval (IR) in the judicial domain is an essential task. With the advent of availability domain-specific data in electronic form and aid of different Artificial intelligence (AI) technologies, automated language processing becomes more comfortable, and hence it becomes feasible for researchers and developers to provide various automated tools to the legal community to reduce human burden. The Competition on Legal Information Extraction/Entailment (COLIEE-2019) run in association with the International Conference on Artificial Intelligence and Law (ICAIL)-2019 has come up with few challenging tasks. The shared defined four sub-tasks (i.e. Task1, Task2, Task3 and Task4), which will be able to provide few automated systems to the judicial system. The paper presents our working note on the experiments carried out as a part of our participation in all the sub-tasks defined in this shared task. We make use of different Information Retrieval(IR) and deep learning based approaches to tackle these problems. We obtain encouraging results in all these four sub-tasks.

(2 min) The Medical Subject Headings (MeSH) thesaurus is a controlled vocabulary widely used in biomedical knowledge systems, particularly for semantic indexing of scientific literature. As the MeSH hierarchy evolves through annual version updates, some new descriptors are introduced that were not previously available. This paper explores the conceptual provenance of these new descriptors. In particular, we investigate whether such new descriptors have been previously covered by older descriptors and what is their current relation to them. To this end, we propose a framework to categorize new descriptors based on their current relation to older descriptors. Based on the proposed classification scheme, we quantify, analyse and present the different types of new descriptors introduced in MeSH during the last fifteen years. The results show that only about 25% of new MeSH descriptors correspond to new emerging concepts, whereas the rest were previously covered by one or more existing descriptors, either implicitly or explicitly. Most of them were covered by a single existing descriptor and they usually end up as descendants of it in the current hierarchy, gradually leading towards a more fine-grained MeSH vocabulary. These insights about the dynamics of the thesaurus are useful for the retrospective study of scientific articles annotated with MeSH, but could also be used to inform the policy of updating the thesaurus in the future.

(2 min) In this paper, we illustrate how to fine-tune the entire Retrieval Augment Generation (RAG) architecture in an end-to-end manner. We highlighted the main engineering challenges that needed to be addressed to achieve this objective. We also compare how end-to-end RAG architecture outperforms the original RAG architecture for the task of question answering. We have open-sourced our implementation in the HuggingFace Transformers library.

(2 min) As job markets worldwide have become more competitive and applicant selection criteria have become more opaque, and different (and sometimes contradictory) information and advice is available for job seekers wishing to progress in their careers, it has never been more difficult to determine which factors in a r\'esum\'e most effectively help career progression. In this work we present a novel, large scale dataset of over half a million r\'esum\'es with preliminary analysis to begin to answer empirically which factors help or hurt people wishing to transition to more senior roles as they progress in their career. We find that previous experience forms the most important factor, outweighing other aspects of human capital, and find which language factors in a r\'esum\'e have significant effects. This lays the groundwork for future inquiry in career trajectories using large scale data analysis and natural language processing techniques.

(2 min) Objective: The objective of this study is to develop a deep learning pipeline to detect signals on dietary supplement-related adverse events (DS AEs) from Twitter. Material and Methods: We obtained 247,807 tweets ranging from 2012 to 2018 that mentioned both DS and AE. We annotated biomedical entities and relations on 2,000 randomly selected tweets. For the concept extraction task, we compared the performance of traditional word embeddings with SVM, CRF and LSTM-CRF classifiers to BERT models. For the relation extraction task, we compared GloVe vectors with CNN classifiers to BERT models. We chose the best performing models in each task to assemble an end-to-end deep learning pipeline to detect DS AE signals and compared the results to the known DS AEs from a DS knowledge base (i.e., iDISK). Results: In both tasks, the BERT-based models outperformed traditional word embeddings. The best performing concept extraction model is the BioBERT model that can identify supplement, symptom, and body organ entities with F1-scores of 0.8646, 0.8497, and 0.7104, respectively. The best performing relation extraction model is the BERT model that can identify purpose and AE relations with F1-scores of 0.8335 and 0.7538, respectively. The end-to-end pipeline was able to extract DS indication and DS AEs with an F1-score of 0.7459 and 0,7414, respectively. Comparing to the iDISK, we could find both known and novel DS-AEs. Conclusion: We have demonstrated the feasibility of detecting DS AE signals from Twitter with a BioBERT-based deep learning pipeline.

(2 min) The emergence of digital technologies such as smartphones in healthcare applications have demonstrated the possibility of developing rich, continuous, and objective measures of multiple sclerosis (MS) disability that can be administered remotely and out-of-clinic. In this work, deep convolutional neural networks (DCNN) applied to smartphone inertial sensor data were shown to better distinguish healthy from MS participant ambulation, compared to standard Support Vector Machine (SVM) feature-based methodologies. To overcome the typical limitations associated with remotely generated health data, such as low subject numbers, sparsity, and heterogeneous data, a transfer learning (TL) model from similar large open-source datasets was proposed. Our TL framework utilised the ambulatory information learned on Human Activity Recognition (HAR) tasks collected from similar smartphone-based sensor data. A lack of transparency of "black-box" deep networks remains one of the largest stumbling blocks to the wider acceptance of deep learning for clinical applications. Ensuing work therefore aimed to visualise DCNN decisions attributed by relevance heatmaps using Layer-Wise Relevance Propagation (LRP). Through the LRP framework, the patterns captured from smartphone-based inertial sensor data that were reflective of those who are healthy versus persons with MS (PwMS) could begin to be established and understood. Interpretations suggested that cadence-based measures, gait speed, and ambulation-related signal perturbations were distinct characteristics that distinguished MS disability from healthy participants. Robust and interpretable outcomes, generated from high-frequency out-of-clinic assessments, could greatly augment the current in-clinic assessment picture for PwMS, to inform better disease management techniques, and enable the development of better therapeutic interventions.

(2 min) Understanding the nutritional content of food from visual data is a challenging computer vision problem, with the potential to have a positive and widespread impact on public health. Studies in this area are limited to existing datasets in the field that lack sufficient diversity or labels required for training models with nutritional understanding capability. We introduce Nutrition5k, a novel dataset of 5k diverse, real world food dishes with corresponding video streams, depth images, component weights, and high accuracy nutritional content annotation. We demonstrate the potential of this dataset by training a computer vision algorithm capable of predicting the caloric and macronutrient values of a complex, real world dish at an accuracy that outperforms professional nutritionists. Further we present a baseline for incorporating depth sensor data to improve nutrition predictions. We will publicly release Nutrition5k in the hope that it will accelerate innovation in the space of nutritional understanding.

(2 min) This paper presents a batch-wise density-based clustering approach for local outlier detection in massive-scale datasets. Unlike the well-known traditional algorithms, which assume that all the data is memory-resident, our proposed method is scalable and processes the input data chunk-by-chunk within the confines of a limited memory buffer. A temporary clustering model is built at the first phase; then, it is gradually updated by analyzing consecutive memory loads of points. Subsequently, at the end of scalable clustering, the approximate structure of the original clusters is obtained. Finally, by another scan of the entire dataset and using a suitable criterion, an outlying score is assigned to each object called SDCOR (Scalable Density-based Clustering Outlierness Ratio). Evaluations on real-life and synthetic datasets demonstrate that the proposed method has a low linear time complexity and is more effective and efficient compared to best-known conventional density-based methods, which need to load all data into the memory; and also, to some fast distance-based methods, which can perform on data resident in the disk.

(2 min) Partial differential equations (PDEs) are concise and understandable representations of domain knowledge, which are essential for deepening our understanding of physical processes and predicting future responses. However, the PDEs of many real-world problems are uncertain, which calls for PDE discovery. We propose the symbolic genetic algorithm (SGA-PDE) to discover open-form PDEs directly from data without prior knowledge about the equation structure. SGA-PDE focuses on the representation and optimization of PDE. Firstly, SGA-PDE uses symbolic mathematics to realize the flexible representation of any given PDE, transforms a PDE into a forest, and converts each function term into a binary tree. Secondly, SGA-PDE adopts a specially designed genetic algorithm to efficiently optimize the binary trees by iteratively updating the tree topology and node attributes. The SGA-PDE is gradient-free, which is a desirable characteristic in PDE discovery since it is difficult to obtain the gradient between the PDE loss and the PDE structure. In the experiment, SGA-PDE not only successfully discovered nonlinear Burgers' equation, Korteweg-de Vries (KdV) equation, and Chafee-Infante equation, but also handled PDEs with fractional structure and compound functions that cannot be solved by conventional PDE discovery methods.

(2 min) We study the statistical theory of offline reinforcement learning (RL) with deep ReLU network function approximation. We analyze a variant of fitted-Q iteration (FQI) algorithm under a new dynamic condition that we call Besov dynamic closure, which encompasses the conditions from prior analyses for deep neural network function approximation. Under Besov dynamic closure, we prove that the FQI-type algorithm enjoys the sample complexity of $\tilde{\mathcal{O}}\left( \kappa^{1 + d/\alpha} \cdot \epsilon^{-2 - 2d/\alpha} \right)$ where $\kappa$ is a distribution shift measure, $d$ is the dimensionality of the state-action space, $\alpha$ is the (possibly fractional) smoothness parameter of the underlying MDP, and $\epsilon$ is a user-specified precision. This is an improvement over the sample complexity of $\tilde{\mathcal{O}}\left( K \cdot \kappa^{2 + d/\alpha} \cdot \epsilon^{-2 - d/\alpha} \right)$ in the prior result [Yang et al., 2019] where $K$ is an algorithmic iteration number which is arbitrarily large in practice. Importantly, our sample complexity is obtained under the new general dynamic condition and a data-dependent structure where the latter is either ignored in prior algorithms or improperly handled by prior analyses. This is the first comprehensive analysis for offline RL with deep ReLU network function approximation under a general setting.

(2 min) Volumen with the Late-Breaking Abstracts submitted to the Evo* 2021 Conference, held online from 7 to 9 of April 2021. These papers present ongoing research and preliminary results investigating on the application of different approaches of Bioinspired Methods (mainly Evolutionary Computation) to different problems, most of them real world ones.

(2 min) The general method of graph coarsening or graph reduction has been a remarkably useful and ubiquitous tool in scientific computing and it is now just starting to have a similar impact in machine learning. The goal of this paper is to take a broad look into coarsening techniques that have been successfully deployed in scientific computing and see how similar principles are finding their way in more recent applications related to machine learning. In scientific computing, coarsening plays a central role in algebraic multigrid methods as well as the related class of multilevel incomplete LU factorizations. In machine learning, graph coarsening goes under various names, e.g., graph downsampling or graph reduction. Its goal in most cases is to replace some original graph by one which has fewer nodes, but whose structure and characteristics are similar to those of the original graph. As will be seen, a common strategy in these methods is to rely on spectral properties to define the coarse graph.

(2 min) In this work, we address the problem of 3D object detection from point cloud data in real time. For autonomous vehicles to work, it is very important for the perception component to detect the real world objects with both high accuracy and fast inference. We propose a novel neural network architecture along with the training and optimization details for detecting 3D objects in point cloud data. We compare the results with different backbone architectures including the standard ones like VGG, ResNet, Inception with our backbone. Also we present the optimization and ablation studies including designing an efficient anchor. We use the Kitti 3D Birds Eye View dataset for benchmarking and validating our results. Our work surpasses the state of the art in this domain both in terms of average precision and speed running at > 30 FPS. This makes it a feasible option to be deployed in real time applications including self driving cars.

(2 min) We consider an adversarially-trained version of the nonnegative matrix factorization, a popular latent dimensionality reduction technique. In our formulation, an attacker adds an arbitrary matrix of bounded norm to the given data matrix. We design efficient algorithms inspired by adversarial training to optimize for dictionary and coefficient matrices with enhanced generalization abilities. Extensive simulations on synthetic and benchmark datasets demonstrate the superior predictive performance on matrix completion tasks of our proposed method compared to state-of-the-art competitors, including other variants of adversarial nonnegative matrix factorization.

(2 min) It is well known that deep neural networks (DNNs) are vulnerable to adversarial attacks, which are implemented by adding crafted perturbations onto benign examples. Min-max robust optimization based adversarial training can provide a notion of security against adversarial attacks. However, adversarial robustness requires a significantly larger capacity of the network than that for the natural training with only benign examples. This paper proposes a framework of concurrent adversarial training and weight pruning that enables model compression while still preserving the adversarial robustness and essentially tackles the dilemma of adversarial training. Furthermore, this work studies two hypotheses about weight pruning in the conventional setting and finds that weight pruning is essential for reducing the network model size in the adversarial setting, training a small model from scratch even with inherited initialization from the large model cannot achieve both adversarial robustness and high standard accuracy. Code is available at https://github.com/yeshaokai/Robustness-Aware-Pruning-ADMM.

(2 min) Deep learning and data-driven approaches have shown great potential in scientific domains. The promise of data-driven techniques relies on the availability of a large volume of high-quality training datasets. Due to the high cost of obtaining data through expensive physical experiments, instruments, and simulations, data augmentation techniques for scientific applications have emerged as a new direction for obtaining scientific data recently. However, existing data augmentation techniques originating from computer vision, yield physically unacceptable data samples that are not helpful for the domain problems that we are interested in. In this paper, we develop new physics-informed data augmentation techniques based on convolutional neural networks. Specifically, our generative models leverage different physics knowledge (such as governing equations, observable perception, and physics phenomena) to improve the quality of the synthetic data. To validate the effectiveness of our data augmentation techniques, we apply them to solve a subsurface seismic full-waveform inversion using simulated CO$_2$ leakage data. Our interest is to invert for subsurface velocity models associated with very small CO$_2$ leakage. We validate the performance of our methods using comprehensive numerical tests. Via comparison and analysis, we show that data-driven seismic imaging can be significantly enhanced by using our physics-informed data augmentation techniques. Particularly, the imaging quality has been improved by 15% in test scenarios of general-sized leakage and 17% in small-sized leakage when using an augmented training set obtained with our techniques.

(2 min) In multi-label classification, where a single example may be associated with several class labels at the same time, the ability to model dependencies between labels is considered crucial to effectively optimize non-decomposable evaluation measures, such as the Subset 0/1 loss. The gradient boosting framework provides a well-studied foundation for learning models that are specifically tailored to such a loss function and recent research attests the ability to achieve high predictive accuracy in the multi-label setting. The utilization of second-order derivatives, as used by many recent boosting approaches, helps to guide the minimization of non-decomposable losses, due to the information about pairs of labels it incorporates into the optimization process. On the downside, this comes with high computational costs, even if the number of labels is small. In this work, we address the computational bottleneck of such approach -- the need to solve a system of linear equations -- by integrating a novel approximation technique into the boosting procedure. Based on the derivatives computed during training, we dynamically group the labels into a predefined number of bins to impose an upper bound on the dimensionality of the linear system. Our experiments, using an existing rule-based algorithm, suggest that this may boost the speed of training, without any significant loss in predictive performance.

(2 min) The non-stationary nature of data streams strongly challenges traditional machine learning techniques. Although some solutions have been proposed to extend traditional machine learning techniques for handling data streams, these approaches either require an initial label set or rely on specialized design parameters. The overlap among classes and the labeling of data streams constitute other major challenges for classifying data streams. In this paper, we proposed a clustering-based data stream classification framework to handle non-stationary data streams without utilizing an initial label set. A density-based stream clustering procedure is used to capture novel concepts with a dynamic threshold and an effective active label querying strategy is introduced to continuously learn the new concepts from the data streams. The sub-cluster structure of each cluster is explored to handle the overlap among classes. Experimental results and quantitative comparison studies reveal that the proposed method provides statistically better or comparable performance than the existing methods.

(2 min) Tensor network methods have been a key ingredient of advances in condensed matter physics and have recently sparked interest in the machine learning community for their ability to compactly represent very high-dimensional objects. Tensor network methods can for example be used to efficiently learn linear models in exponentially large feature spaces [Stoudenmire and Schwab, 2016]. In this work, we derive upper and lower bounds on the VC dimension and pseudo-dimension of a large class of tensor network models for classification, regression and completion. Our upper bounds hold for linear models parameterized by arbitrary tensor network structures, and we derive lower bounds for common tensor decomposition models~(CP, Tensor Train, Tensor Ring and Tucker) showing the tightness of our general upper bound. These results are used to derive a generalization bound which can be applied to classification with low rank matrices as well as linear classifiers based on any of the commonly used tensor decomposition models. As a corollary of our results, we obtain a bound on the VC dimension of the matrix product state classifier introduced in [Stoudenmire and Schwab, 2016] as a function of the so-called bond dimension~(i.e. tensor train rank), which answers an open problem listed by Cirac, Garre-Rubio and P\'erez-Garc\'ia in [Cirac et al., 2019].

(2 min) Background: Colonoscopy remains the gold-standard screening for colorectal cancer. However, significant miss rates for polyps have been reported, particularly when there are multiple small adenomas. This presents an opportunity to leverage computer-aided systems to support clinicians and reduce the number of polyps missed. Method: In this work we introduce the Focus U-Net, a novel dual attention-gated deep neural network, which combines efficient spatial and channel-based attention into a single Focus Gate module to encourage selective learning of polyp features. The Focus U-Net further incorporates short-range skip connections and deep supervision. Furthermore, we introduce the Hybrid Focal loss, a new compound loss function based on the Focal loss and Focal Tversky loss, to handle class-imbalanced image segmentation. For our experiments, we selected five public datasets containing images of polyps obtained during optical colonoscopy: CVC-ClinicDB, Kvasir-SEG, CVC-ColonDB, ETIS-Larib PolypDB and EndoScene test set. To evaluate model performance, we use the Dice similarity coefficient (DSC) and Intersection over Union (IoU) metrics. Results: Our model achieves state-of-the-art results for both CVC-ClinicDB and Kvasir-SEG, with a mean DSC of 0.941 and 0.910, respectively. When evaluated on a combination of five public polyp datasets, our model similarly achieves state-of-the-art results with a mean DSC of 0.878 and mean IoU of 0.809, a 14% and 15% improvement over the previous state-of-the-art results of 0.768 and 0.702, respectively. Conclusions: This study shows the potential for deep learning to provide fast and accurate polyp segmentation results for use during colonoscopy. The Focus U-Net may be adapted for future use in newer non-invasive screening and more broadly to other biomedical image segmentation tasks involving class imbalance and requiring efficiency.

(2 min) Reinforcement learning (RL) aims to find an optimal policy by interaction with an environment. Consequently, learning complex behavior requires a vast number of samples, which can be prohibitive in practice. Nevertheless, instead of systematically reasoning and actively choosing informative samples, policy gradients for local search are often obtained from random perturbations. These random samples yield high variance estimates and hence are sub-optimal in terms of sample complexity. Actively selecting informative samples is at the core of Bayesian optimization, which constructs a probabilistic surrogate of the objective from past samples to reason about informative subsequent ones. In this paper, we propose to join both worlds. We develop an algorithm utilizing a probabilistic model of the objective function and its gradient. Based on the model, the algorithm decides where to query a noisy zeroth-order oracle to improve the gradient estimates. The resulting algorithm is a novel type of policy search method, which we compare to existing black-box algorithms. The comparison reveals improved sample complexity and reduced variance in extensive empirical evaluations on synthetic objectives. Further, we highlight the benefits of active sampling on popular RL benchmarks.

(2 min) The success of deep reinforcement learning (DRL) is due to the power of learning a representation that is suitable for the underlying exploration and exploitation task. However, existing provable reinforcement learning algorithms with linear function approximation often assume the feature representation is known and fixed. In order to understand how representation learning can improve the efficiency of RL, we study representation learning for a class of low-rank Markov Decision Processes (MDPs) where the transition kernel can be represented in a bilinear form. We propose a provably efficient algorithm called ReLEX that can simultaneously learn the representation and perform exploration. We show that ReLEX always performs no worse than a state-of-the-art algorithm without representation learning, and will be strictly better in terms of sample efficiency if the function class of representations enjoys a certain mild "coverage'' property over the whole state-action space.

(2 min) In this paper, we present a novel neuroevolutionary method to identify the architecture and hyperparameters of convolutional autoencoders. Remarkably, we used a hypervolume indicator in the context of neural architecture search for autoencoders, for the first time to our current knowledge. Results show that images were compressed by a factor of more than 10, while still retaining enough information to achieve image classification for the majority of the tasks. Thus, this new approach can be used to speed up the AutoML pipeline for image compression.

(2 min) Data augmentation has been successfully used in many areas of deep-learning to significantly improve model performance. Typically data augmentation simulates realistic variations in data in order to increase the apparent diversity of the training-set. However, for opcode-based malware analysis, where deep learning methods are already achieving state of the art performance, it is not immediately clear how to apply data augmentation. In this paper we study different methods of data augmentation starting with basic methods using fixed transformations and moving to methods that adapt to the data. We propose a novel data augmentation method based on using an opcode embedding layer within the network and its corresponding opcode embedding matrix to perform adaptive data augmentation during training. To the best of our knowledge this is the first paper to carry out a systematic study of different augmentation methods applied to opcode sequence based malware classification.

(2 min) In this work, we present a novel neural network to generate high resolution images. We replace the decoder of VAE with a discriminator while using the encoder as it is. The encoder is fed data from a normal distribution while the generator is fed from a gaussian distribution. The combination from both is given to a discriminator which tells whether the generated image is correct or not. We evaluate our network on 3 different datasets: MNIST, LSUN and CelebA dataset. Our network beats the previous state of the art using MMD, SSIM, log likelihood, reconstruction error, ELBO and KL divergence as the evaluation metrics while generating much sharper images. This work is potentially very exciting as we are able to combine the advantages of generative models and inference models in a principled bayesian manner.

(2 min) We aim to develop clustering models to obtain behavioral representations from continuous multimodal mobile sensing data towards relapse prediction tasks. The identified clusters could represent different routine behavioral trends related to daily living of patients as well as atypical behavioral trends associated with impending relapse. We used the mobile sensing data obtained in the CrossCheck project for our analysis. Continuous data from six different mobile sensing-based modalities (e.g. ambient light, sound/conversation, acceleration etc.) obtained from a total of 63 schizophrenia patients, each monitored for up to a year, were used for the clustering models and relapse prediction evaluation. Two clustering models, Gaussian Mixture Model (GMM) and Partition Around Medoids (PAM), were used to obtain behavioral representations from the mobile sensing data. The features obtained from the clustering models were used to train and evaluate a personalized relapse prediction model using Balanced Random Forest. The personalization was done by identifying optimal features for a given patient based on a personalization subset consisting of other patients who are of similar age. The clusters identified using the GMM and PAM models were found to represent different behavioral patterns (such as clusters representing sedentary days, active but with low communications days, etc.). Significant changes near the relapse periods were seen in the obtained behavioral representation features from the clustering models. The clustering model based features, together with other features characterizing the mobile sensing data, resulted in an F2 score of 0.24 for the relapse prediction task in a leave-one-patient-out evaluation setting. This obtained F2 score is significantly higher than a random classification baseline with an average F2 score of 0.042.

(2 min) Algorithmic recourse aims to provide actionable recommendations to individuals to obtain a more favourable outcome from an automated decision-making system. As it involves reasoning about interventions performed in the physical world, recourse is fundamentally a causal problem. Existing methods compute the effect of recourse actions using a causal model learnt from data under the assumption of no hidden confounding and modelling assumptions such as additive noise. Building on the seminal work of Balke and Pearl (1994), we propose an alternative approach for discrete random variables which relaxes these assumptions and allows for unobserved confounding and arbitrary structural equations. The proposed approach only requires specification of the causal graph and confounding structure and bounds the expected counterfactual effect of recourse actions. If the lower bound is above a certain threshold, i.e., on the other side of the decision boundary, recourse is guaranteed in expectation.

(2 min) We study fast algorithms for statistical regression problems under the strong contamination model, where the goal is to approximately optimize a generalized linear model (GLM) given adversarially corrupted samples. Prior works in this line of research were based on the robust gradient descent framework of Prasad et. al., a first-order method using biased gradient queries, or the Sever framework of Diakonikolas et. al., an iterative outlier-removal method calling a stationary point finder. We present nearly-linear time algorithms for robust regression problems with improved runtime or estimation guarantees compared to the state-of-the-art. For the general case of smooth GLMs (e.g. logistic regression), we show that the robust gradient descent framework of Prasad et. al. can be accelerated, and show our algorithm extends to optimizing the Moreau envelopes of Lipschitz GLMs (e.g. support vector machines), answering several open questions in the literature. For the well-studied case of robust linear regression, we present an alternative approach obtaining improved estimation rates over prior nearly-linear time algorithms. Interestingly, our method starts with an identifiability proof introduced in the context of the sum-of-squares algorithm of Bakshi and Prasad, which achieved optimal error rates while requiring large polynomial runtime and sample complexity. We reinterpret their proof within the Sever framework and obtain a dramatically faster and more sample-efficient algorithm under fewer distributional assumptions.

(2 min) The H-measure is a classifier performance measure which takes into account the context of application without requiring a rigid value of relative misclassification costs to be set. Since its introduction in 2009 it has become widely adopted. This paper answers various queries which users have raised since its introduction, including questions about its interpretation, the choice of a weighting function, whether it is strictly proper, and its coherence, and relates the measure to other work.

(2 min) Conditional value-at-risk (CVaR) and value-at-risk (VaR) are popular tail-risk measures in finance and insurance industries as well as in highly reliable, safety-critical uncertain environments where often the underlying probability distributions are heavy-tailed. We use the multi-armed bandit best-arm identification framework and consider the problem of identifying the arm from amongst finitely many that has the smallest CVaR, VaR, or weighted sum of CVaR and mean. The latter captures the risk-return trade-off common in finance. Our main contribution is an optimal $\delta$-correct algorithm that acts on general arms, including heavy-tailed distributions, and matches the lower bound on the expected number of samples needed, asymptotically (as $\delta$ approaches $0$). The algorithm requires solving a non-convex optimization problem in the space of probability measures, that requires delicate analysis. En-route, we develop new non-asymptotic empirical likelihood-based concentration inequalities for tail-risk measures which are tighter than those for popular truncation-based empirical estimators.

(2 min) \textit{Stochastic gradient descent} (SGD) is of fundamental importance in deep learning. Despite its simplicity, elucidating its efficacy remains challenging. Conventionally, the success of SGD is attributed to the \textit{stochastic gradient noise} (SGN) incurred in the training process. Based on this general consensus, SGD is frequently treated and analyzed as the Euler-Maruyama discretization of a \textit{stochastic differential equation} (SDE) driven by either Brownian or L\'evy stable motion. In this study, we argue that SGN is neither Gaussian nor stable. Instead, inspired by the long-time correlation emerging in SGN series, we propose that SGD can be viewed as a discretization of an SDE driven by \textit{fractional Brownian motion} (FBM). Accordingly, the different convergence behavior of SGD dynamics is well grounded. Moreover, the first passage time of an SDE driven by FBM is approximately derived. This indicates a lower escaping rate for a larger Hurst parameter, and thus SGD stays longer in flat minima. This happens to coincide with the well-known phenomenon that SGD favors flat minima that generalize well. Four groups of experiments are conducted to validate our conjecture, and it is demonstrated that long-range memory effects persist across various model architectures, datasets, and training strategies. Our study opens up a new perspective and may contribute to a better understanding of SGD.

(2 min) Speech production is a dynamic procedure, which involved multi human organs including the tongue, jaw and lips. Modeling the dynamics of the vocal tract deformation is a fundamental problem to understand the speech, which is the most common way for human daily communication. Researchers employ several sensory streams to describe the process simultaneously, which are incontrovertibly statistically related to other streams. In this paper, we address the following question: given an observable image sequences of lips, can we picture the corresponding tongue motion. We formulated this problem as the self-supervised learning problem, and employ the two-stream convolutional network and long-short memory network for the learning task, with the attention mechanism. We evaluate the performance of the proposed method by leveraging the unlabeled lip videos to predict an upcoming ultrasound tongue image sequence. The results show that our model is able to generate images that close to the real ultrasound tongue images, and results in the matching between two imaging modalities.

(2 min) We investigate the possibility of forcing a self-supervised model trained using a contrastive predictive loss to extract slowly varying latent representations. Rather than producing individual predictions for each of the future representations, the model emits a sequence of predictions shorter than that of the upcoming representations to which they will be aligned. In this way, the prediction network solves a simpler task of predicting the next symbols, but not their exact timing, while the encoding network is trained to produce piece-wise constant latent codes. We evaluate the model on a speech coding task and demonstrate that the proposed Aligned Contrastive Predictive Coding (ACPC) leads to higher linear phone prediction accuracy and lower ABX error rates, while being slightly faster to train due to the reduced number of prediction heads.

(2 min) The recent rise of privacy concerns has led researchers to devise methods for private neural inference -- where inferences are made directly on encrypted data, never seeing inputs. The primary challenge facing private inference is that computing on encrypted data levies an impractically-high latency penalty, stemming mostly from non-linear operators like ReLU. Enabling practical and private inference requires new optimization methods that minimize network ReLU counts while preserving accuracy. This paper proposes DeepReDuce: a set of optimizations for the judicious removal of ReLUs to reduce private inference latency. The key insight is that not all ReLUs contribute equally to accuracy. We leverage this insight to drop, or remove, ReLUs from classic networks to significantly reduce inference latency and maintain high accuracy. Given a target network, DeepReDuce outputs a Pareto frontier of networks that tradeoff the number of ReLUs and accuracy. Compared to the state-of-the-art for private inference DeepReDuce improves accuracy and reduces ReLU count by up to 3.5% (iso-ReLU count) and 3.5$\times$ (iso-accuracy), respectively.

(2 min) Subspace tracking (ST) with missing data (ST-miss) or outliers (Robust ST) or both (Robust ST-miss) has been extensively studied in the last many years. This work provides a new simple algorithm and guarantee for both ST with missing data (ST-miss) and RST-miss. Unlike past work on this topic, the algorithm is much simpler (uses fewer parameters) and the guarantee does not make the artificial assumption of piecewise constant subspace change, although it still handles that setting. Secondly, we extend our approach and its analysis to provably solving these problems when the raw data is federated and when the over-air data communication modality is used for information exchange between the $K$ peer nodes and the center.

(2 min) Dysfluencies and variations in speech pronunciation can severely degrade speech recognition performance, and for many individuals with moderate-to-severe speech disorders, voice operated systems do not work. Current speech recognition systems are trained primarily with data from fluent speakers and as a consequence do not generalize well to speech with dysfluencies such as sound or word repetitions, sound prolongations, or audible blocks. The focus of this work is on quantitative analysis of a consumer speech recognition system on individuals who stutter and production-oriented approaches for improving performance for common voice assistant tasks (i.e., "what is the weather?"). At baseline, this system introduces a significant number of insertion and substitution errors resulting in intended speech Word Error Rates (isWER) that are 13.64\% worse (absolute) for individuals with fluency disorders. We show that by simply tuning the decoding parameters in an existing hybrid speech recognition system one can improve isWER by 24\% (relative) for individuals with fluency disorders. Tuning these parameters translates to 3.6\% better domain recognition and 1.7\% better intent recognition relative to the default setup for the 18 study participants across all stuttering severities.

(2 min) We study the stochastic shortest path problem with adversarial costs and known transition, and show that the minimax regret is $\widetilde{O}(\sqrt{DT^\star K})$ and $\widetilde{O}(\sqrt{DT^\star SA K})$ for the full-information setting and the bandit feedback setting respectively, where $D$ is the diameter, $T^\star$ is the expected hitting time of the optimal policy, $S$ is the number of states, $A$ is the number of actions, and $K$ is the number of episodes. Our results significantly improve upon the existing work of (Rosenberg and Mansour, 2020) which only considers the full-information setting and achieves suboptimal regret. Our work is also the first to consider bandit feedback with adversarial costs. Our algorithms are built on top of the Online Mirror Descent framework with a variety of new techniques that might be of independent interest, including an improved multi-scale expert algorithm, a reduction from general stochastic shortest path to a special loop-free case, a skewed occupancy measure space, and a novel correction term added to the cost estimators. Interestingly, the last two elements reduce the variance of the learner via positive bias and the variance of the optimal policy via negative bias respectively, and having them simultaneously is critical for obtaining the optimal high-probability bound in the bandit feedback setting.

(2 min) Optimization in Deep Learning is mainly guided by vague intuitions and strong assumptions, with a limited understanding how and why these work in practice. To shed more light on this, our work provides some deeper understandings of how SGD behaves by empirically analyzing the trajectory taken by SGD from a line search perspective. Specifically, a costly quantitative analysis of the full-batch loss along SGD trajectories from common used models trained on a subset of CIFAR-10 is performed. Our core results include that the full-batch loss along lines in update step direction is highly parabolically. Further on, we show that there exists a learning rate with which SGD always performs almost exact line searches on the full-batch loss. Finally, we provide a different perspective why increasing the batch size has almost the same effect as decreasing the learning rate by the same factor.

(2 min) With the prevalence of online social networking sites (OSNs) and mobile devices, people are increasingly reliant on a variety of OSNs for keeping in touch with family and friends, and using it as a source of information. For example, a user might utilise multiple OSNs for different purposes, such as using Flickr to share holiday pictures with family and friends, and Twitter to post short messages about their thoughts. Identifying the same user across multiple OSNs is an important task as this allows us to understand the usage patterns of users among different OSNs, make recommendations when a user registers for a new OSN, and various other useful applications. To address this problem, we proposed an algorithm based on the multilayer perceptron using various types of features, namely: (i) user profile, such as name, location, description; (ii) temporal distribution of user generated content; and (iii) embedding based on user name, real name and description. Using a Twitter and Flickr dataset of users and their posting activities, we perform an empirical study on how these features affect the performance of user identification across the two OSNs and discuss our main findings based on the different features.

(2 min) Topic modeling is an unsupervised method for revealing the hidden semantic structure of a corpus. It has been increasingly widely adopted as a tool in the social sciences, including political science, digital humanities and sociological research in general. One desirable property of topic models is to allow users to find topics describing a specific aspect of the corpus. A possible solution is to incorporate domain-specific knowledge into topic modeling, but this requires a specification from domain experts. We propose a novel query-driven topic model that allows users to specify a simple query in words or phrases and return query-related topics, thus avoiding tedious work from domain experts. Our proposed approach is particularly attractive when the user-specified query has a low occurrence in a text corpus, making it difficult for traditional topic models built on word cooccurrence patterns to identify relevant topics. Experimental results demonstrate the effectiveness of our model in comparison with both classical topic models and neural topic models.

(2 min) Recent work demonstrated that flow-based invertible neural networks are promising tools for solving ambiguous inverse problems. Following up on this, we investigate how ten invertible architectures and related models fare on two intuitive, low-dimensional benchmark problems, obtaining the best results with coupling layers and simple autoencoders. We hope that our initial efforts inspire other researchers to evaluate their invertible architectures in the same setting and put forth additional benchmarks, so our evaluation may eventually grow into an official community challenge.

(2 min) Voice anti-spoofing aims at classifying a given utterance either as a bonafide human sample, or a spoofing attack (e.g. synthetic or replayed sample). Many anti-spoofing methods have been proposed but most of them fail to generalize across domains (corpora) -- and we do not know \emph{why}. We outline a novel interpretative framework for gauging the impact of data quality upon anti-spoofing performance. Our within- and between-domain experiments pool data from seven public corpora and three anti-spoofing methods based on Gaussian mixture and convolutive neural network models. We assess the impacts of long-term spectral information, speaker population (through x-vector speaker embeddings), signal-to-noise ratio, and selected voice quality features.

(2 min) Cross-silo federated learning (FL) is a distributed learning approach where clients train a global model cooperatively while keeping their local data private. Different from cross-device FL, clients in cross-silo FL are usually organizations or companies which may execute multiple cross-silo FL processes repeatedly due to their time-varying local data sets, and aim to optimize their long-term benefits by selfishly choosing their participation levels. While there has been some work on incentivizing clients to join FL, the analysis of the long-term selfish participation behaviors of clients in cross-silo FL remains largely unexplored. In this paper, we analyze the selfish participation behaviors of heterogeneous clients in cross-silo FL. Specifically, we model the long-term selfish participation behaviors of clients as an infinitely repeated game, with the stage game being a selfish participation game in one cross-silo FL process (SPFL). For the stage game SPFL, we derive the unique Nash equilibrium (NE), and propose a distributed algorithm for each client to calculate its equilibrium participation strategy. For the long-term interactions among clients, we derive a cooperative strategy for clients which minimizes the number of free riders while increasing the amount of local data for model training. We show that enforced by a punishment strategy, such a cooperative strategy is a SPNE of the infinitely repeated game, under which some clients who are free riders at the NE of the stage game choose to be (partial) contributors. We further propose an algorithm to calculate the optimal SPNE which minimizes the number of free riders while maximizing the amount of local data for model training. Simulation results show that our proposed cooperative strategy at the optimal SPNE can effectively reduce the number of free riders and increase the amount of local data for model training.

(2 min) Data assimilation (DA) is widely used to combine physical knowledge and observations. It is nowadays commonly used in geosciences to perform parametric calibration. In a context of climate change, old calibrations can not necessarily be used for new scenarios. This raises the question of DA computational cost, as costly physics-based numerical models need to be reanalyzed. Reduction and metamodelling represent therefore interesting perspectives, for example proposed in recent contributions as hybridization between ensemble and variational methods, to combine their advantages (efficiency, non-linear framework). They are however often based on Monte Carlo (MC) type sampling, which often requires considerable increase of the ensemble size for better efficiency, therefore representing a computational burden in ensemble-based methods as well. To address these issues, two methods to replace the complex model by a surrogate are proposed and confronted : (i) PODEn3DVAR directly inspired from PODEn4DVAR, relies on an ensemble-based joint parameter-state Proper Orthogonal Decomposition (POD), which provides a linear metamodel ; (ii) POD-PCE-3DVAR, where the model states are POD reduced then learned using Polynomial Chaos Expansion (PCE), resulting in a non-linear metamodel. Both metamodels allow to write an approximate cost function whose minimum can be analytically computed, or deduced by a gradient descent at negligible cost. Furthermore, adapted metamodelling error covariance matrix is given for POD-PCE-3DVAR, allowing to substantially improve the metamodel-based DA analysis. Proposed methods are confronted on a twin experiment, and compared to classical 3DVAR on a measurement-based problem. Results are promising, in particular superior with POD-PCE-3DVAR, showing good convergence to classical 3DVAR and robustness to noise.

(2 min) Spiking Neural Networks (SNNs) have been attached great importance due to their biological plausibility and high energy-efficiency on neuromorphic chips. As these chips are usually resource-constrained, the compression of SNNs is thus crucial along the road of practical use of SNNs. Most existing methods directly apply pruning approaches in artificial neural networks (ANNs) to SNNs, which ignore the difference between ANNs and SNNs, thus limiting the performance of the pruned SNNs. Besides, these methods are only suitable for shallow SNNs. In this paper, inspired by synaptogenesis and synapse elimination in the neural system, we propose gradient rewiring (Grad R), a joint learning algorithm of connectivity and weight for SNNs, that enables us to seamlessly optimize network structure without retraining. Our key innovation is to redefine the gradient to a new synaptic parameter, allowing better exploration of network structures by taking full advantage of the competition between pruning and regrowth of connections. The experimental results show that the proposed method achieves minimal loss of SNNs' performance on MNIST and CIFAR-10 dataset so far. Moreover, it reaches a $\sim$3.5% accuracy loss under unprecedented 0.73% connectivity, which reveals remarkable structure refining capability in SNNs. Our work suggests that there exists extremely high redundancy in deep SNNs. Our codes are available at https://github.com/Yanqi-Chen/Gradient-Rewiring.

(2 min) Explaining Graph Neural Networks predictions to end users of AI applications in easily understandable terms remains an unsolved problem. In particular, we do not have well developed methods for automatically evaluating explanations, in ways that are closer to how users consume those explanations. Based on recent application trends and our own experiences in real world problems, we propose automatic evaluation approaches for GNN Explanations.

(2 min) Machine-generated artworks are now part of the contemporary art scene: they are attracting significant investments and they are presented in exhibitions together with those created by human artists. These artworks are mainly based on generative deep learning techniques. Also given their success, several legal problems arise when working with these techniques. In this article we consider a set of key questions in the area of generative deep learning for the arts. Is it possible to use copyrighted works as training set for generative models? How do we legally store their copies in order to perform the training process? And then, who (if someone) will own the copyright on the generated data? We try to answer these questions considering the law in force in both US and EU and the future alternatives, trying to define a set of guidelines for artists and developers working on deep learning generated art.

(2 min) Active Learning is essential for more label-efficient deep learning. Bayesian Active Learning has focused on BALD, which reduces model parameter uncertainty. However, we show that BALD gets stuck on out-of-distribution or junk data that is not relevant for the task. We examine a novel *Expected Predictive Information Gain (EPIG)* to deal with distribution shifts of the pool set. EPIG reduces the uncertainty of *predictions* on an unlabelled *evaluation set* sampled from the test data distribution whose distribution might be different to the pool set distribution. Based on this, our new EPIG-BALD acquisition function for Bayesian Neural Networks selects samples to improve the performance on the test data distribution instead of selecting samples that reduce model uncertainty everywhere, including for out-of-distribution regions with low density in the test data distribution. Our method outperforms state-of-the-art Bayesian active learning methods on high-dimensional datasets and avoids out-of-distribution junk data in cases where current state-of-the-art methods fail.

(2 min) We introduce the framework of continuous-depth graph neural networks (GNNs). Neural graph differential equations (Neural GDEs) are formalized as the counterpart to GNNs where the input-output relationship is determined by a continuum of GNN layers, blending discrete topological structures and differential equations. The proposed framework is shown to be compatible with static GNN models and is extended to dynamic and stochastic settings through hybrid dynamical system theory. Here, Neural GDEs improve performance by exploiting the underlying dynamics geometry, further introducing the ability to accommodate irregularly sampled data. Results prove the effectiveness of the proposed models across applications, such as traffic forecasting or prediction in genetic regulatory networks.

(2 min) Machine learning methods are widely used in the natural sciences to model and predict physical systems from observation data. Yet, they are often used as poorly understood "black boxes," disregarding existing mathematical structure and invariants of the problem. Recently, the proposal of Hamiltonian Neural Networks (HNNs) took a first step towards a unified "gray box" approach, using physical insight to improve performance for Hamiltonian systems. In this paper, we explore a significantly improved training method for HNNs, exploiting the symplectic structure of Hamiltonian systems with a different loss function. This frees the loss from an artificial lower bound. We mathematically guarantee the existence of an exact Hamiltonian function which the HNN can learn. This allows us to prove and numerically analyze the errors made by HNNs which, in turn, renders them fully explainable. Finally, we present a novel post-training correction to obtain the true Hamiltonian only from discretized observation data, up to an arbitrary order.

(2 min) The quest for determinism in machine learning has disproportionately focused on characterizing the impact of noise introduced by algorithmic design choices. In this work, we address a less well understood and studied question: how does our choice of tooling introduce randomness to deep neural network training. We conduct large scale experiments across different types of hardware, accelerators, state of art networks, and open-source datasets, to characterize how tooling choices contribute to the level of non-determinism in a system, the impact of said non-determinism, and the cost of eliminating different sources of noise. Our findings are surprising, and suggest that the impact of non-determinism in nuanced. While top-line metrics such as top-1 accuracy are not noticeably impacted, model performance on certain parts of the data distribution is far more sensitive to the introduction of randomness. Our results suggest that deterministic tooling is critical for AI safety. However, we also find that the cost of ensuring determinism varies dramatically between neural network architectures and hardware types, e.g., with overhead up to $746\%$, $241\%$, and $196\%$ on a spectrum of widely used GPU accelerator architectures, relative to non-deterministic training. The source code used in this paper is available at https://github.com/usyd-fsalab/NeuralNetworkRandomness.

(2 min) In this work, we extend standard neural networks by building upon an assumption that neuronal activations correspond to the angle of a complex number lying on the unit circle, or 'phasor.' Each layer in such a network produces new activations by taking a weighted superposition of the previous layer's phases and calculating the new phase value. This generalized architecture allows models to reach high accuracy and carries the singular advantage that mathematically equivalent versions of the network can be executed with or without regard to a temporal variable. Importantly, the value of a phase angle in the temporal domain can be sparsely represented by a periodically repeating series of delta functions or 'spikes'. We demonstrate the atemporal training of a phasor network on standard deep learning tasks and show that these networks can then be executed in either the traditional atemporal domain or spiking temporal domain with no conversion step needed. This provides a novel basis for constructing deep networkswhich operate via temporal, spike-based calculations suitable for neuromorphic computing hardware.

(2 min) Hierarchical Agglomerative Clustering (HAC) algorithms are extensively utilized in modern data science, and seek to partition the dataset into clusters while generating a hierarchical relationship between the data samples. HAC algorithms are employed in many applications, such as biology, natural language processing, and recommender systems. Thus, it is imperative to ensure that these algorithms are fair -- even if the dataset contains biases against certain protected groups, the cluster outputs generated should not discriminate against samples from any of these groups. However, recent work in clustering fairness has mostly focused on center-based clustering algorithms, such as k-median and k-means clustering. In this paper, we propose fair algorithms for performing HAC that enforce fairness constraints 1) irrespective of the distance linkage criteria used, 2) generalize to any natural measures of clustering fairness for HAC, 3) work for multiple protected groups, and 4) have competitive running times to vanilla HAC. Through extensive experiments on multiple real-world UCI datasets, we show that our proposed algorithm finds fairer clusterings compared to vanilla HAC as well as other state-of-the-art fair clustering approaches.

(2 min) Semi-supervised learning has been gaining attention as it allows for performing image analysis tasks such as classification with limited labeled data. Some popular algorithms using Generative Adversarial Networks (GANs) for semi-supervised classification share a single architecture for classification and discrimination. However, this may require a model to converge to a separate data distribution for each task, which may reduce overall performance. While progress in semi-supervised learning has been made, less addressed are small-scale, fully-supervised tasks where even unlabeled data is unavailable and unattainable. We therefore, propose a novel GAN model namely External Classifier GAN (EC-GAN), that utilizes GANs and semi-supervised algorithms to improve classification in fully-supervised regimes. Our method leverages a GAN to generate artificial data used to supplement supervised classification. More specifically, we attach an external classifier, hence the name EC-GAN, to the GAN's generator, as opposed to sharing an architecture with the discriminator. Our experiments demonstrate that EC-GAN's performance is comparable to the shared architecture method, far superior to the standard data augmentation and regularization-based approach, and effective on a small, realistic dataset.

(2 min) We introduce a new framework for unifying and systematizing the performance analysis of first-order black-box optimization algorithms for unconstrained convex minimization. The low-cost iteration complexity enjoyed by first-order algorithms renders them particularly relevant for applications in machine learning and large-scale data analysis. Relying on sum-of-squares (SOS) optimization, we introduce a hierarchy of semidefinite programs that give increasingly better convergence bounds for higher levels of the hierarchy. Alluding to the power of the SOS hierarchy, we show that the (dual of the) first level corresponds to the Performance Estimation Problem (PEP) introduced by Drori and Teboulle [Math. Program., 145(1):451--482, 2014], a powerful framework for determining convergence rates of first-order optimization algorithms. Consequently, many results obtained within the PEP framework can be reinterpreted as degree-1 SOS proofs, and thus, the SOS framework provides a promising new approach for certifying improved rates of convergence by means of higher-order SOS certificates. To determine analytical rate bounds, in this work we use the first level of the SOS hierarchy and derive new result{s} for noisy gradient descent with inexact line search methods (Armijo, Wolfe, and Goldstein).

(2 min) The growth in Remote Patient Monitoring (RPM) services using wearable and non-wearable Internet of Medical Things (IoMT) promises to improve the quality of diagnosis and facilitate timely treatment for a gamut of medical conditions. At the same time, the proliferation of IoMT devices increases the potential for malicious activities that can lead to catastrophic results including theft of personal information, data breach, and compromised medical devices, putting human lives at risk. IoMT devices generate tremendous amount of data that reflect user behavior patterns including both personal and day-to-day social activities along with daily routine health monitoring. In this context, there are possibilities of anomalies generated due to various reasons including unexpected user behavior, faulty sensor, or abnormal values from malicious/compromised devices. To address this problem, there is an imminent need to develop a framework for securing the smart health care infrastructure to identify and mitigate anomalies. In this paper, we present an anomaly detection model for RPM utilizing IoMT and smart home devices. We propose Hidden Markov Model (HMM) based anomaly detection that analyzes normal user behavior in the context of RPM comprising both smart home and smart health devices, and identifies anomalous user behavior. We design a testbed with multiple IoMT devices and home sensors to collect data and use the HMM model to train using network and user behavioral data. Proposed HMM based anomaly detection model achieved over 98% accuracy in identifying the anomalies in the context of RPM.

(2 min) We resolve the long-standing "impossible tuning" issue for the classic expert problem and show that, it is in fact possible to achieve regret $O\left(\sqrt{(\ln d)\sum_t \ell_{t,i}^2}\right)$ simultaneously for all expert $i$ in a $T$-round $d$-expert problem where $\ell_{t,i}$ is the loss for expert $i$ in round $t$. Our algorithm is based on the Mirror Descent framework with a correction term and a weighted entropy regularizer. While natural, the algorithm has not been studied before and requires a careful analysis. We also generalize the bound to $O\left(\sqrt{(\ln d)\sum_t (\ell_{t,i}-m_{t,i})^2}\right)$ for any prediction vector $m_t$ that the learner receives, and recover or improve many existing results by choosing different $m_t$. Furthermore, we use the same framework to create a master algorithm that combines a set of base algorithms and learns the best one with little overhead. The new guarantee of our master allows us to derive many new results for both the expert problem and more generally Online Linear Optimization.

(2 min) Enabling effective and efficient machine learning (ML) over large-scale graph data (e.g., graphs with billions of edges) can have a huge impact on both industrial and scientific applications. However, community efforts to advance large-scale graph ML have been severely limited by the lack of a suitable public benchmark. For KDD Cup 2021, we present OGB Large-Scale Challenge (OGB-LSC), a collection of three real-world datasets for advancing the state-of-the-art in large-scale graph ML. OGB-LSC provides graph datasets that are orders of magnitude larger than existing ones and covers three core graph learning tasks -- link prediction, graph regression, and node classification. Furthermore, OGB-LSC provides dedicated baseline experiments, scaling up expressive graph ML models to the massive datasets. We show that the expressive models significantly outperform simple scalable baselines, indicating an opportunity for dedicated efforts to further improve graph ML at scale. Our datasets and baseline code are released and maintained as part of our OGB initiative (Hu et al., 2020). We hope OGB-LSC at KDD Cup 2021 can empower the community to discover innovative solutions for large-scale graph ML.

(2 min) We study bandits and reinforcement learning (RL) subject to a conservative constraint where the agent is asked to perform at least as well as a given baseline policy. This setting is particular relevant in real-world domains including digital marketing, healthcare, production, finance, etc. For multi-armed bandits, linear bandits and tabular RL, specialized algorithms and theoretical analyses were proposed in previous work. In this paper, we present a unified framework for conservative bandits and RL, in which our core technique is to calculate the necessary and sufficient budget obtained from running the baseline policy. For lower bounds, our framework gives a black-box reduction that turns a certain lower bound in the nonconservative setting into a new lower bound in the conservative setting. We strengthen the existing lower bound for conservative multi-armed bandits and obtain new lower bounds for conservative linear bandits, tabular RL and low-rank MDP. For upper bounds, our framework turns a certain nonconservative upper-confidence-bound (UCB) algorithm into a conservative algorithm with a simple analysis. For multi-armed bandits, linear bandits and tabular RL, our new upper bounds tighten or match existing ones with significantly simpler analyses. We also obtain a new upper bound for conservative low-rank MDP.

(2 min) A graph-inspired classifier, known as Optimum-Path Forest (OPF), has proven to be a state-of-the-art algorithm comparable to Logistic Regressors, Support Vector Machines in a wide variety of tasks. Recently, its Python-based version, denoted as OPFython, has been proposed to provide a more friendly framework and a faster prototyping environment. Nevertheless, Python-based algorithms are slower than their counterpart C-based algorithms, impacting their performance when confronted with large amounts of data. Therefore, this paper proposed a simple yet highly efficient speed up using the Numba package, which accelerates Numpy-based calculations and attempts to increase the algorithm's overall performance. Experimental results showed that the proposed approach achieved better results than the na\"ive Python-based OPF and speeded up its distance measurement calculation.

(2 min) Multiple lines of evidence suggest that predictive models may benefit from algorithmic triage. Under algorithmic triage, a predictive model does not predict all instances but instead defers some of them to human experts. However, the interplay between the prediction accuracy of the model and the human experts under algorithmic triage is not well understood. In this work, we start by formally characterizing under which circumstances a predictive model may benefit from algorithmic triage. In doing so, we also demonstrate that models trained for full automation may be suboptimal under triage. Then, given any model and desired level of triage, we show that the optimal triage policy is a deterministic threshold rule in which triage decisions are derived deterministically by thresholding the difference between the model and human errors on a per-instance level. Building upon these results, we introduce a practical gradient-based algorithm that is guaranteed to find a sequence of triage policies and predictive models of increasing performance. Experiments on a wide variety of supervised learning tasks using synthetic and real data from two important applications -- content moderation and scientific discovery -- illustrate our theoretical results and show that the models and triage policies provided by our gradient-based algorithm outperform those provided by several competitive baselines.

(2 min) Recent advances in protecting node privacy on graph data and attacking graph neural networks (GNNs) gain much attention. The eye does not bring these two essential tasks together yet. Imagine an adversary can utilize the powerful GNNs to infer users' private labels in a social network. How can we adversarially defend against such privacy attacks while maintaining the utility of perturbed graphs? In this work, we propose a novel research task, adversarial defenses against GNN-based privacy attacks, and present a graph perturbation-based approach, NetFense, to achieve the goal. NetFense can simultaneously keep graph data unnoticeability (i.e., having limited changes on the graph structure), maintain the prediction confidence of targeted label classification (i.e., preserving data utility), and reduce the prediction confidence of private label classification (i.e., protecting the privacy of nodes). Experiments conducted on single- and multiple-target perturbations using three real graph data exhibit that the perturbed graphs by NetFense can effectively maintain data utility (i.e., model unnoticeability) on targeted label classification and significantly decrease the prediction confidence of private label classification (i.e., privacy protection). Extensive studies also bring several insights, such as the flexibility of NetFense, preserving local neighborhoods in data unnoticeability, and better privacy protection for high-degree nodes.

(2 min) Deep neural networks have reached very high accuracy on object detection but their success hinges on large amounts of labeled data. To reduce the dependency on labels, various active-learning strategies have been proposed, typically based on the confidence of the detector. However, these methods are biased towards best-performing classes and can lead to acquired datasets that are not good representatives of the data in the testing set. In this work, we propose a unified framework for active learning, that considers both the uncertainty and the robustness of the detector, ensuring that the network performs accurately in all classes. Furthermore, our method is able to pseudo-label the very confident predictions, suppressing a potential distribution drift while further boosting the performance of the model. Experiments show that our method comprehensively outperforms a wide range of active-learning methods on PASCAL VOC07+12 and MS-COCO, having up to a 7.7% relative improvement, or up to 82% reduction in labeling cost.

(2 min) Self-supervised learning (especially contrastive learning) has attracted great interest due to its tremendous potentials in learning discriminative representations in an unsupervised manner. Despite the acknowledged successes, existing contrastive learning methods suffer from very low learning efficiency, e.g., taking about ten times more training epochs than supervised learning for comparable recognition accuracy. In this paper, we discover two contradictory phenomena in contrastive learning that we call under-clustering and over-clustering problems, which are major obstacles to learning efficiency. Under-clustering means that the model cannot efficiently learn to discover the dissimilarity between inter-class samples when the negative sample pairs for contrastive learning are insufficient to differentiate all the actual object categories. Over-clustering implies that the model cannot efficiently learn the feature representation from excessive negative sample pairs, which enforces the model to over-cluster samples of the same actual categories into different clusters. To simultaneously overcome these two problems, we propose a novel self-supervised learning framework using a median triplet loss. Precisely, we employ a triplet loss tending to maximize the relative distance between the positive pair and negative pairs to address the under-clustering problem; and we construct the negative pair by selecting the negative sample of a median similarity score from all negative samples to avoid the over-clustering problem, guaranteed by the Bernoulli Distribution model. We extensively evaluate our proposed framework in several large-scale benchmarks (e.g., ImageNet, SYSU-30k, and COCO). The results demonstrate the superior performance (e.g., the learning efficiency) of our model over the latest state-of-the-art methods by a clear margin. Codes available at: https://github.com/wanggrun/triplet.

(2 min) Networks have been widely used to represent the relations between objects such as academic networks and social networks, and learning embedding for networks has thus garnered plenty of research attention. Self-supervised network representation learning aims at extracting node embedding without external supervision. Recently, maximizing the mutual information between the local node embedding and the global summary (e.g. Deep Graph Infomax, or DGI for short) has shown promising results on many downstream tasks such as node classification. However, there are two major limitations of DGI. Firstly, DGI merely considers the extrinsic supervision signal (i.e., the mutual information between node embedding and global summary) while ignores the intrinsic signal (i.e., the mutual dependence between node embedding and node attributes). Secondly, nodes in a real-world network are usually connected by multiple edges with different relations, while DGI does not fully explore the various relations among nodes. To address the above-mentioned problems, we propose a novel framework, called High-order Deep Multiplex Infomax (HDMI), for learning node embedding on multiplex networks in a self-supervised way. To be more specific, we first design a joint supervision signal containing both extrinsic and intrinsic mutual information by high-order mutual information, and we propose a High-order Deep Infomax (HDI) to optimize the proposed supervision signal. Then we propose an attention based fusion module to combine node embedding from different layers of the multiplex network. Finally, we evaluate the proposed HDMI on various downstream tasks such as unsupervised clustering and supervised classification. The experimental results show that HDMI achieves state-of-the-art performance on these tasks.

(2 min) We propose a new stochastic gradient method that uses recorded past loss values to reduce the variance. Our method can be interpreted as a new stochastic variant of the Polyak Stepsize that converges globally without assuming interpolation. Our method introduces auxiliary variables, one for each data point, that track the loss value for each data point. We provide a global convergence theory for our method by showing that it can be interpreted as a special variant of online SGD. The new method only stores a single scalar per data point, opening up new applications for variance reduction where memory is the bottleneck.

(2 min) Differentiable Architecture Search (DARTS) is a recently proposed neural architecture search (NAS) method based on a differentiable relaxation. Due to its success, numerous variants analyzing and improving parts of the DARTS framework have recently been proposed. By considering the problem as a constrained bilevel optimization, we propose and analyze three improvements to architectural weight competition, update scheduling, and regularization towards discretization. First, we introduce a new approach to the activation of architecture weights, which prevents confounding competition within an edge and allows for fair comparison across edges to aid in discretization. Next, we propose a dynamic schedule based on per-minibatch network information to make architecture updates more informed. Finally, we consider two regularizations, based on proximity to discretization and the Alternating Directions Method of Multipliers (ADMM) algorithm, to promote early discretization. Our results show that this new activation scheme reduces final architecture size and the regularizations improve reliability in search results while maintaining comparable performance to state-of-the-art in NAS, especially when used with our new dynamic informed schedule.

(2 min) In the environment of fair lending laws and the General Data Protection Regulation (GDPR), the ability to explain a model's prediction is of paramount importance. High quality explanations are the first step in assessing fairness. Counterfactuals are valuable tools for explainability. They provide actionable, comprehensible explanations for the individual who is subject to decisions made from the prediction. It is important to find a baseline for producing them. We propose a simple method for generating counterfactuals by using gradient descent to search in the latent space of an autoencoder and benchmark our method against approaches that search for counterfactuals in feature space. Additionally, we implement metrics to concretely evaluate the quality of the counterfactuals. We show that latent space counterfactual generation strikes a balance between the speed of basic feature gradient descent methods and the sparseness and authenticity of counterfactuals generated by more complex feature space oriented techniques.

(2 min) The goal of bayesian approach used in variational inference is to minimize the KL divergence between variational distribution and unknown posterior distribution. This is done by maximizing the Evidence Lower Bound (ELBO). A neural network is used to parametrize these distributions using Stochastic Gradient Descent. This work extends the work done by others by deriving the variational inference models. We show how SGD can be applied on bayesian neural networks by gradient estimation techniques. For validation, we have tested our model on 5 UCI datasets and the metrics chosen for evaluation are Root Mean Square Error (RMSE) error and negative log likelihood. Our work considerably beats the previous state of the art approaches for regression using bayesian neural networks.

(2 min) UK Biobank (UKB) is conducting a large-scale study of more than half a million volunteers, collecting health-related information on genetics, lifestyle, blood biochemistry, and more. Medical imaging furthermore targets 100,000 subjects, with 70,000 follow-up sessions, enabling measurements of organs, muscle, and body composition. With up to 170,000 mounting MR images, various methodologies are accordingly engaged in large-scale image analysis. This work presents an experimental inference engine that can automatically predict a comprehensive profile of subject metadata from UKB neck-to-knee body MRI. In cross-validation, it accurately inferred baseline characteristics such as age, height, weight, and sex, but also emulated measurements of body composition by DXA, organ volumes, and abstract properties like grip strength, pulse rate, and type 2 diabetic status (AUC: 0.866). The proposed system can automatically analyze thousands of subjects within hours and provide individual confidence intervals. The underlying methodology is based on convolutional neural networks for image-based mean-variance regression on two-dimensional representations of the MRI data. This work aims to make the proposed system available for free to researchers, who can use it to obtain fast and fully-automated estimates of 72 different measurements immediately upon release of new UK Biobank image data.

(2 min) Modern pre-trained language models are mostly built upon backbones stacking self-attention and feed-forward layers in an interleaved order. In this paper, beyond this stereotyped layer pattern, we aim to improve pre-trained models by exploiting layer variety from two aspects: the layer type set and the layer order. Specifically, besides the original self-attention and feed-forward layers, we introduce convolution into the layer type set, which is experimentally found beneficial to pre-trained models. Furthermore, beyond the original interleaved order, we explore more layer orders to discover more powerful architectures. However, the introduced layer variety leads to a large architecture space of more than billions of candidates, while training a single candidate model from scratch already requires huge computation cost, making it not affordable to search such a space by directly training large amounts of candidate models. To solve this problem, we first pre-train a supernet from which the weights of all candidate models can be inherited, and then adopt an evolutionary algorithm guided by pre-training accuracy to find the optimal architecture. Extensive experiments show that LV-BERT model obtained by our method outperforms BERT and its variants on various downstream tasks. For example, LV-BERT-small achieves 78.8 on the GLUE testing set, 1.8 higher than the strong baseline ELECTRA-small.

(2 min) In this paper, we aim to solve the high dimensional stochastic optimal control problem from the view of the stochastic maximum principle via deep learning. By introducing the extended Hamiltonian system which is essentially an FBSDE with a maximum condition, we reformulate the original control problem as a new one. Three algorithms are proposed to solve the new control problem. Numerical results for different examples demonstrate the effectiveness of our proposed algorithms, especially in high dimensional cases. And an important application of this method is to calculate the sub-linear expectations, which correspond to a kind of fully nonlinear PDEs.

(2 min) Self-training is an effective approach to semi-supervised learning. The key idea is to let the learner itself iteratively generate "pseudo-supervision" for unlabeled instances based on its current hypothesis. In combination with consistency regularization, pseudo-labeling has shown promising performance in various domains, for example in computer vision. To account for the hypothetical nature of the pseudo-labels, these are commonly provided in the form of probability distributions. Still, one may argue that even a probability distribution represents an excessive level of informedness, as it suggests that the learner precisely knows the ground-truth conditional probabilities. In our approach, we therefore allow the learner to label instances in the form of credal sets, that is, sets of (candidate) probability distributions. Thanks to this increased expressiveness, the learner is able to represent uncertainty and a lack of knowledge in a more flexible and more faithful manner. To learn from weakly labeled data of that kind, we leverage methods that have recently been proposed in the realm of so-called superset learning. In an exhaustive empirical evaluation, we compare our methodology to state-of-the-art self-supervision approaches, showing competitive to superior performance especially in low-label scenarios incorporating a high degree of uncertainty.

(2 min) Off-policy learning allows us to learn about possible policies of behavior from experience generated by a different behavior policy. Temporal difference (TD) learning algorithms can become unstable when combined with function approximation and off-policy sampling - this is known as the ''deadly triad''. Emphatic temporal difference (ETD($\lambda$)) algorithm ensures convergence in the linear case by appropriately weighting the TD($\lambda$) updates. In this paper, we extend the use of emphatic methods to deep reinforcement learning agents. We show that naively adapting ETD($\lambda$) to popular deep reinforcement learning algorithms, which use forward view multi-step returns, results in poor performance. We then derive new emphatic algorithms for use in the context of such algorithms, and we demonstrate that they provide noticeable benefits in small problems designed to highlight the instability of TD methods. Finally, we observed improved performance when applying these algorithms at scale on classic Atari games from the Arcade Learning Environment.

(2 min) Multiple organ failure (MOF) is a severe syndrome with a high mortality rate among Intensive Care Unit (ICU) patients. Early and precise detection is critical for clinicians to make timely decisions. An essential challenge in applying machine learning models to electronic health records (EHRs) is the pervasiveness of missing values. Most existing imputation methods are involved in the data preprocessing phase, failing to capture the relationship between data and outcome for downstream predictions. In this paper, we propose classifier-guided generative adversarial imputation networks Classifier-GAIN) for MOF prediction to bridge this gap, by incorporating both observed data and label information. Specifically, the classifier takes imputed values from the generator(imputer) to predict task outcomes and provides additional supervision signals to the generator by joint training. The classifier-guide generator imputes missing values with label-awareness during training, improving the classifier's performance during inference. We conduct extensive experiments showing that our approach consistently outperforms classical and state-of-art neural baselines across a range of missing data scenarios and evaluation metrics.

(2 min) We consider stochastic optimization with delayed gradients where, at each time step $t$, the algorithm makes an update using a stale stochastic gradient from step $t - d_t$ for some arbitrary delay $d_t$. This setting abstracts asynchronous distributed optimization where a central server receives gradient updates computed by worker machines. These machines can experience computation and communication loads that might vary significantly over time. In the general non-convex smooth optimization setting, we give a simple and efficient algorithm that requires $O( \sigma^2/\epsilon^4 + \tau/\epsilon^2 )$ steps for finding an $\epsilon$-stationary point $x$, where $\tau$ is the \emph{average} delay $\smash{\frac{1}{T}\sum_{t=1}^T d_t}$ and $\sigma^2$ is the variance of the stochastic gradients. This improves over previous work, which showed that stochastic gradient decent achieves the same rate but with respect to the \emph{maximal} delay $\max_{t} d_t$, that can be significantly larger than the average delay especially in heterogeneous distributed systems. Our experiments demonstrate the efficacy and robustness of our algorithm in cases where the delay distribution is skewed or heavy-tailed.

(2 min) Gender-based crime is one of the most concerning scourges of contemporary society. Governments worldwide have invested lots of economic and human resources to radically eliminate this threat. Despite these efforts, providing accurate predictions of the risk that a victim of gender violence has of being attacked again is still a very hard open problem. The development of new methods for issuing accurate, fair and quick predictions would allow police forces to select the most appropriate measures to prevent recidivism. In this work, we propose to apply Machine Learning (ML) techniques to create models that accurately predict the recidivism risk of a gender-violence offender. The relevance of the contribution of this work is threefold: (i) the proposed ML method outperforms the preexisting risk assessment algorithm based on classical statistical techniques, (ii) the study has been conducted through an official specific-purpose database with more than 40,000 reports of gender violence, and (iii) two new quality measures are proposed for assessing the effective police protection that a model supplies and the overload in the invested resources that it generates. Additionally, we propose a hybrid model that combines the statistical prediction methods with the ML method, permitting authorities to implement a smooth transition from the preexisting model to the ML-based model. This hybrid nature enables a decision-making process to optimally balance between the efficiency of the police system and aggressiveness of the protection measures taken.

(2 min) This paper presents a novel fingerprinting methodology for the Intellectual Property protection of generative models. Prior solutions for discriminative models usually adopt adversarial examples as the fingerprints, which give anomalous inference behaviors and prediction results. Hence, these methods are not stealthy and can be easily recognized by the adversary. Our approach leverages the invisible backdoor technique to overcome the above limitation. Specifically, we design verification samples, whose model outputs look normal but can trigger a backdoor classifier to make abnormal predictions. We propose a new backdoor embedding approach with Unique-Triplet Loss and fine-grained categorization to enhance the effectiveness of our fingerprints. Extensive evaluations show that this solution can outperform other strategies with higher robustness, uniqueness and stealthiness for various GAN models.

(2 min) We study reinforcement learning (RL) with linear function approximation. Existing algorithms for this problem only have high-probability regret and/or Probably Approximately Correct (PAC) sample complexity guarantees, which cannot guarantee the convergence to the optimal policy. In this paper, in order to overcome the limitation of existing algorithms, we propose a new algorithm called FLUTE, which enjoys uniform-PAC convergence to the optimal policy with high probability. The uniform-PAC guarantee is the strongest possible guarantee for reinforcement learning in the literature, which can directly imply both PAC and high probability regret bounds, making our algorithm superior to all existing algorithms with linear function approximation. At the core of our algorithm is a novel minimax value function estimator and a multi-level partition scheme to select the training samples from historical observations. Both of these techniques are new and of independent interest.

(2 min) When tuning the architecture and hyperparameters of large machine learning models for on-device deployment, it is desirable to understand the optimal trade-offs between on-device latency and model accuracy. In this work, we leverage recent methodological advances in Bayesian optimization over high-dimensional search spaces and multi-objective Bayesian optimization to efficiently explore these trade-offs for a production-scale on-device natural language understanding model at Facebook.

(2 min) In this work, we consider the problem of learning a feed-forward neural network (NN) controller to safely steer an arbitrarily shaped planar robot in a compact and obstacle-occluded workspace. Unlike existing methods that depend strongly on the density of data points close to the boundary of the safe state space to train NN controllers with closed-loop safety guarantees, we propose an approach that lifts such assumptions on the data that are hard to satisfy in practice and instead allows for graceful safety violations, i.e., of a bounded magnitude that can be spatially controlled. To do so, we employ reachability analysis methods to encapsulate safety constraints in the training process. Specifically, to obtain a computationally efficient over-approximation of the forward reachable set of the closed-loop system, we partition the robot's state space into cells and adaptively subdivide the cells that contain states which may escape the safe set under the trained control law. To do so, we first design appropriate under- and over-approximations of the robot's footprint to adaptively subdivide the configuration space into cells. Then, using the overlap between each cell's forward reachable set and the set of infeasible robot configurations as a measure for safety violations, we introduce penalty terms into the loss function that penalize this overlap in the training process. As a result, our method can learn a safe vector field for the closed-loop system and, at the same time, provide numerical worst-case bounds on safety violation over the whole configuration space, defined by the overlap between the over-approximation of the forward reachable set of the closed-loop system and the set of unsafe states. Moreover, it can control the tradeoff between computational complexity and tightness of these bounds. Finally, we provide a simulation study that verifies the efficacy of the proposed scheme.

(2 min) Conventional image classifiers are trained by randomly sampling mini-batches of images. To achieve state-of-the-art performance, practitioners use sophisticated data augmentation schemes to expand the amount of training data available for sampling. In contrast, meta-learning algorithms sample support data, query data, and tasks on each training step. In this complex sampling scenario, data augmentation can be used not only to expand the number of images available per class, but also to generate entirely new classes/tasks. We systematically dissect the meta-learning pipeline and investigate the distinct ways in which data augmentation can be integrated at both the image and class levels. Our proposed meta-specific data augmentation significantly improves the performance of meta-learners on few-shot classification benchmarks.

(2 min) Early detection of suicidal ideation in depressed individuals can allow for adequate medical attention and support, which in many cases is life-saving. Recent NLP research focuses on classifying, from a given piece of text, if an individual is suicidal or clinically healthy. However, there have been no major attempts to differentiate between depression and suicidal ideation, which is an important clinical challenge. Due to the scarce availability of EHR data, suicide notes, or other similar verified sources, web query data has emerged as a promising alternative. Online sources, such as Reddit, allow for anonymity that prompts honest disclosure of symptoms, making it a plausible source even in a clinical setting. However, these online datasets also result in lower performance, which can be attributed to the inherent noise in web-scraped labels, which necessitates a noise-removal process. Thus, we propose SDCNL, a suicide versus depression classification method through a deep learning approach. We utilize online content from Reddit to train our algorithm, and to verify and correct noisy labels, we propose a novel unsupervised label correction method which, unlike previous work, does not require prior noise distribution information. Our extensive experimentation with multiple deep word embedding models and classifiers display the strong performance of the method in anew, challenging classification application. We make our code and dataset available at https://github.com/ayaanzhaque/SDCNL

(2 min) We consider a class-incremental semantic segmentation (CISS) problem. While some recently proposed algorithms utilized variants of knowledge distillation (KD) technique to tackle the problem, they only partially addressed the key additional challenges in CISS that causes the catastrophic forgetting; i.e., the semantic drift of the background class and multi-label prediction issue. To better address these challenges, we propose a new method, dubbed as SSUL-M (Semantic Segmentation with Unknown Label with Memory), by carefully combining several techniques tailored for semantic segmentation. More specifically, we make three main contributions; (1) modeling unknown class within the background class to help learning future classes (help plasticity), (2) freezing backbone network and past classifiers with binary cross-entropy loss and pseudo-labeling to overcome catastrophic forgetting (help stability), and (3) utilizing tiny exemplar memory for the first time in CISS to improve both plasticity and stability. As a result, we show our method achieves significantly better performance than the recent state-of-the-art baselines on the standard benchmark datasets. Furthermore, we justify our contributions with thorough and extensive ablation analyses and discuss different natures of the CISS problem compared to the standard class-incremental learning for classification.

(2 min) Graph representation learning is a fundamental problem for modeling relational data and benefits a number of downstream applications. Traditional Bayesian-based graph models and recent deep learning based GNN either suffer from impracticability or lack interpretability, thus combined models for undirected graphs have been proposed to overcome the weaknesses. As a large portion of real-world graphs are directed graphs (of which undirected graphs are special cases), in this paper, we propose a Deep Latent Space Model (DLSM) for directed graphs to incorporate the traditional latent variable based generative model into deep learning frameworks. Our proposed model consists of a graph convolutional network (GCN) encoder and a stochastic decoder, which are layer-wise connected by a hierarchical variational auto-encoder architecture. By specifically modeling the degree heterogeneity using node random factors, our model possesses better interpretability in both community structure and degree heterogeneity. For fast inference, the stochastic gradient variational Bayes (SGVB) is adopted using a non-iterative recognition model, which is much more scalable than traditional MCMC-based methods. The experiments on real-world datasets show that the proposed model achieves the state-of-the-art performances on both link prediction and community detection tasks while learning interpretable node embeddings. The source code is available at https://github.com/upperr/DLSM.

(2 min) We present a novel deep neural network (DNN) architecture for compressing an image when a correlated image is available as side information only at the decoder. This problem is known as distributed source coding (DSC) in information theory. In particular, we consider a pair of stereo images, which generally have high correlation with each other due to overlapping fields of view, and assume that one image of the pair is to be compressed and transmitted, while the other image is available only at the decoder. In the proposed architecture, the encoder maps the input image to a latent space, quantizes the latent representation, and compresses it using entropy coding. The decoder is trained to extract the Wyner's common information between the input image and the correlated image from the latter. The received latent representation and the locally generated common information are passed through a decoder network to obtain an enhanced reconstruction of the input image. The common information provides a succinct representation of the relevant information at the receiver. We train and demonstrate the effectiveness of the proposed approach on the KITTI dataset of stereo image pairs. Our results show that the proposed architecture is capable of exploiting the decoder-only side information, and outperforms previous work on stereo image compression with decoder side information.

(2 min) The emergence of deep learning has been accompanied by privacy concerns surrounding users' data and service providers' models. We focus on private inference (PI), where the goal is to perform inference on a user's data sample using a service provider's model. Existing PI methods for deep networks enable cryptographically secure inference with little drop in functionality; however, they incur severe latency costs, primarily caused by non-linear network operations (such as ReLUs). This paper presents Sphynx, a ReLU-efficient network design method based on micro-search strategies for convolutional cell design. Sphynx achieves Pareto dominance over all existing private inference methods on CIFAR-100. We also design large-scale networks that support cryptographically private inference on Tiny-ImageNet and ImageNet.

(2 min) In the real world, many tasks require multiple agents to cooperate with each other under the condition of local observations. To solve such problems, many multi-agent reinforcement learning methods based on Centralized Training with Decentralized Execution have been proposed. One representative class of work is value decomposition, which decomposes the global joint Q-value $Q_\text{jt}$ into individual Q-values $Q_a$ to guide individuals' behaviors, e.g. VDN (Value-Decomposition Networks) and QMIX. However, these baselines often ignore the randomness in the situation. We propose MMD-MIX, a method that combines distributional reinforcement learning and value decomposition to alleviate the above weaknesses. Besides, to improve data sampling efficiency, we were inspired by REM (Random Ensemble Mixture) which is a robust RL algorithm to explicitly introduce randomness into the MMD-MIX. The experiments demonstrate that MMD-MIX outperforms prior baselines in the StarCraft Multi-Agent Challenge (SMAC) environment.

(2 min) We propose a novel and effective input transformation based adversarial defense method against gray- and black-box attack, which is computationally efficient and does not require any adversarial training or retraining of a classification model. We first show that a very simple iterative Gaussian smoothing can effectively wash out adversarial noise and achieve substantially high robust accuracy. Based on the observation, we propose Self-Supervised Iterative Contextual Smoothing (SSICS), which aims to reconstruct the original discriminative features from the Gaussian-smoothed image in context-adaptive manner, while still smoothing out the adversarial noise. From the experiments on ImageNet, we show that our SSICS achieves both high standard accuracy and very competitive robust accuracy for the gray- and black-box attacks; e.g., transfer-based PGD-attack and score-based attack. A note-worthy point to stress is that our defense is free of computationally expensive adversarial training, yet, can approach its robust accuracy via input transformation.

(2 min) We study the off-policy evaluation (OPE) problem in reinforcement learning with linear function approximation, which aims to estimate the value function of a target policy based on the offline data collected by a behavior policy. We propose to incorporate the variance information of the value function to improve the sample efficiency of OPE. More specifically, for time-inhomogeneous episodic linear Markov decision processes (MDPs), we propose an algorithm, VA-OPE, which uses the estimated variance of the value function to reweight the Bellman residual in Fitted Q-Iteration. We show that our algorithm achieves a tighter error bound than the best-known result. We also provide a fine-grained characterization of the distribution shift between the behavior policy and the target policy. Extensive numerical experiments corroborate our theory.

(3 min) Common compartmental modeling for COVID-19 is based on a priori knowledge and numerous assumptions. Additionally, they do not systematically incorporate asymptomatic cases. Our study aimed at providing a framework for data-driven approaches, by leveraging the strengths of the grey-box system theory or grey-box identification, known for its robustness in problem solving under partial, incomplete, or uncertain data. Empirical data on confirmed cases and deaths, extracted from an open source repository were used to develop the SEAIRD compartment model. Adjustments were made to fit current knowledge on the COVID-19 behavior. The model was implemented and solved using an Ordinary Differential Equation solver and an optimization tool. A cross-validation technique was applied, and the coefficient of determination $R^2$ was computed in order to evaluate the goodness-of-fit of the model. %to the data. Key epidemiological parameters were finally estimated and we provided the rationale for the construction of SEAIRD model. When applied to Brazil's cases, SEAIRD produced an excellent agreement to the data, with an %coefficient of determination $R^2$ $\geq 90\%$. The probability of COVID-19 transmission was generally high ($\geq 95\%$). On the basis of a 20-day modeling data, the incidence rate of COVID-19 was as low as 3 infected cases per 100,000 exposed persons in Brazil and France. Within the same time frame, the fatality rate of COVID-19 was the highest in France (16.4\%) followed by Brazil (6.9\%), and the lowest in Russia ($\leq 1\%$). SEAIRD represents an asset for modeling infectious diseases in their dynamical stable phase, especially for new viruses when pathophysiology knowledge is very limited.

(2 min) Digitalization leads to data transparency for production systems that we can benefit from with data-driven analysis methods like neural networks. For example, automated anomaly detection enables saving resources and optimizing the production. We study using rarely occurring information about labeled anomalies into Variational Autoencoder neural network structures to overcome information deficits of supervised and unsupervised approaches. This method outperforms all other models in terms of accuracy, precision, and recall. We evaluate the following methods: Principal Component Analysis, Isolation Forest, Classifying Neural Networks, and Variational Autoencoders on seven time series datasets to find the best performing detection methods. We extend this idea to include more infrequently occurring meta information about production processes. This use of sparse labels, both of anomalies or production data, allows to harness any additional information available for increasing anomaly detection performance.

(2 min) The classical Langevin Monte Carlo method looks for i.i.d. samples from a target distribution by descending along the gradient of the target distribution. It is popular partially due to its fast convergence rate. However, the numerical cost is sometimes high because the gradient can be hard to obtain. One approach to eliminate the gradient computation is to employ the concept of "ensemble", where a large number of particles are evolved together so that the neighboring particles provide gradient information to each other. In this article, we discuss two algorithms that integrate the ensemble feature into LMC, and the associated properties. There are two sides of our discovery: 1. By directly surrogating the gradient using the ensemble approximation, we develop Ensemble Langevin Monte Carlo. We show that this method is unstable due to a potentially small denominator that induces high variance. We provide a counterexample to explicitly show this instability. 2. We then change the strategy and enact the ensemble approximation to the gradient only in a constrained manner, to eliminate the unstable points. The algorithm is termed Constrained Ensemble Langevin Monte Carlo. We show that, with a proper tuning, the surrogation takes place often enough to bring the reasonable numerical saving, while the induced error is still low enough for us to maintain the fast convergence rate, up to a controllable discretization and ensemble error. Such combination of ensemble method and LMC shed light on inventing gradient-free algorithms that produce i.i.d. samples almost exponentially fast.

(2 min) In many computer vision classification tasks, class priors at test time often differ from priors on the training set. In the case of such prior shift, classifiers must be adapted correspondingly to maintain close to optimal performance. This paper analyzes methods for adaptation of probabilistic classifiers to new priors and for estimating new priors on an unlabeled test set. We propose a novel method to address a known issue of prior estimation methods based on confusion matrices, where inconsistent estimates of decision probabilities and confusion matrices lead to negative values in the estimated priors. Experiments on fine-grained image classification datasets provide insight into the best practice of prior shift estimation and classifier adaptation and show that the proposed method achieves state-of-the-art results in prior adaptation. Applying the best practice to two tasks with naturally imbalanced priors, learning from web-crawled images and plant species classification, increased the recognition accuracy by 1.1% and 3.4% respectively.

(2 min) We study two-sided matching markets in which one side of the market (the players) does not have a priori knowledge about its preferences for the other side (the arms) and is required to learn its preferences from experience. Also, we assume the players have no direct means of communication. This model extends the standard stochastic multi-armed bandit framework to a decentralized multiple player setting with competition. We introduce a new algorithm for this setting that, over a time horizon $T$, attains $\mathcal{O}(\log(T))$ stable regret when preferences of the arms over players are shared, and $\mathcal{O}(\log(T)^2)$ regret when there are no assumptions on the preferences on either side. Moreover, in the setting where a single player may deviate, we show that the algorithm is incentive compatible whenever the arms' preferences are shared, but not necessarily so when preferences are fully general.

(2 min) Differential equations in general and neural ODEs in particular are an essential technique in continuous-time system identification. While many deterministic learning algorithms have been designed based on numerical integration via the adjoint method, many downstream tasks such as active learning, exploration in reinforcement learning, robust control, or filtering require accurate estimates of predictive uncertainties. In this work, we propose a novel approach towards estimating epistemically uncertain neural ODEs, avoiding the numerical integration bottleneck. Instead of modeling uncertainty in the ODE parameters, we directly model uncertainties in the state space. Our algorithm - distributional gradient matching (DGM) - jointly trains a smoother and a dynamics model and matches their gradients via minimizing a Wasserstein loss. Our experiments show that, compared to traditional approximate inference methods based on numerical integration, our approach is faster to train, faster at predicting previously unseen trajectories, and in the context of neural ODEs, significantly more accurate.

(2 min) Replay strategies are Continual Learning techniques which mitigate catastrophic forgetting by keeping a buffer of patterns from previous experiences, which are interleaved with new data during training. The amount of patterns stored in the buffer is a critical parameter which largely influences the final performance and the memory footprint of the approach. This work introduces Distilled Replay, a novel replay strategy for Continual Learning which is able to mitigate forgetting by keeping a very small buffer (1 pattern per class) of highly informative samples. Distilled Replay builds the buffer through a distillation process which compresses a large dataset into a tiny set of informative examples. We show the effectiveness of our Distilled Replay against popular replay-based strategies on four Continual Learning benchmarks.

(2 min) The stochastic gradient descent (SGD) algorithm is widely used for parameter estimation, especially for huge data sets and online learning. While this recursive algorithm is popular for computation and memory efficiency, quantifying variability and randomness of the solutions has been rarely studied. This paper aims at conducting statistical inference of SGD-based estimates in an online setting. In particular, we propose a fully online estimator for the covariance matrix of averaged SGD iterates (ASGD) only using the iterates from SGD. We formally establish our online estimator's consistency and show that the convergence rate is comparable to offline counterparts. Based on the classic asymptotic normality results of ASGD, we construct asymptotically valid confidence intervals for model parameters. Upon receiving new observations, we can quickly update the covariance matrix estimate and the confidence intervals. This approach fits in an online setting and takes full advantage of SGD: efficiency in computation and memory.

(2 min) The necessity of deep learning for tabular data is still an unanswered question addressed by a large number of research efforts. The recent literature on tabular DL proposes several deep architectures reported to be superior to traditional "shallow" models like Gradient Boosted Decision Trees. However, since existing works often use different benchmarks and tuning protocols, it is unclear if the proposed models universally outperform GBDT. Moreover, the models are often not compared to each other, therefore, it is challenging to identify the best deep model for practitioners. In this work, we start from a thorough review of the main families of DL models recently developed for tabular data. We carefully tune and evaluate them on a wide range of datasets and reveal two significant findings. First, we show that the choice between GBDT and DL models highly depends on data and there is still no universally superior solution. Second, we demonstrate that a simple ResNet-like architecture is a surprisingly effective baseline, which outperforms most of the sophisticated models from the DL literature. Finally, we design a simple adaptation of the Transformer architecture for tabular data that becomes a new strong DL baseline and reduces the gap between GBDT and DL models on datasets where GBDT dominates.

(2 min) In this paper, we study the problem of estimating smooth Generalized Linear Models (GLM) in the Non-interactive Local Differential Privacy (NLDP) model. Different from its classical setting, our model allows the server to access some additional public but unlabeled data. By using Stein's lemma and its variants, we first show that there is an $(\epsilon, \delta)$-NLDP algorithm for GLM (under some mild assumptions), if each data record is i.i.d sampled from some sub-Gaussian distribution with bounded $\ell_1$-norm. Then with high probability, the sample complexity of the public and private data, for the algorithm to achieve an $\alpha$ estimation error (in $\ell_\infty$-norm), is $O(p^2\alpha^{-2})$ and ${O}(p^2\alpha^{-2}\epsilon^{-2})$, respectively, if $\alpha$ is not too small ({\em i.e.,} $\alpha\geq \Omega(\frac{1}{\sqrt{p}})$), where $p$ is the dimensionality of the data. This is a significant improvement over the previously known quasi-polynomial (in $\alpha$) or exponential (in $p$) complexity of GLM with no public data. Also, our algorithm can answer multiple (at most $\exp(O(p))$) GLM queries with the same sample complexities as in the one GLM query case with at least constant probability. We then extend our idea to the non-linear regression problem and show a similar phenomenon for it. Finally, we demonstrate the effectiveness of our algorithms through experiments on both synthetic and real world datasets. To our best knowledge, this is the first paper showing the existence of efficient and effective algorithms for GLM and non-linear regression in the NLDP model with public unlabeled data.

(2 min) There have been many recent advances on provably efficient Reinforcement Learning (RL) in problems with rich observation spaces. However, all these works share a strong realizability assumption about the optimal value function of the true MDP. Such realizability assumptions are often too strong to hold in practice. In this work, we consider the more realistic setting of agnostic RL with rich observation spaces and a fixed class of policies $\Pi$ that may not contain any near-optimal policy. We provide an algorithm for this setting whose error is bounded in terms of the rank $d$ of the underlying MDP. Specifically, our algorithm enjoys a sample complexity bound of $\widetilde{O}\left((H^{4d} K^{3d} \log |\Pi|)/\epsilon^2\right)$ where $H$ is the length of episodes, $K$ is the number of actions and $\epsilon>0$ is the desired sub-optimality. We also provide a nearly matching lower bound for this agnostic setting that shows that the exponential dependence on rank is unavoidable, without further assumptions.

(2 min) Fairness-aware learning aims at constructing classifiers that not only make accurate predictions, but do not discriminate against specific groups. It is a fast-growing area of machine learning with far-reaching societal impact. However, existing fair learning methods are vulnerable to accidental or malicious artifacts in the training data, which can cause them to unknowingly produce unfair classifiers. In this work we address the problem of fair learning from unreliable training data in the robust multisource setting, where the available training data comes from multiple sources, a fraction of which might be not representative of the true data distribution. We introduce FLEA, a filtering-based algorithm that allows the learning system to identify and suppress those data sources that would have a negative impact on fairness or accuracy if they were used for training. We show the effectiveness of our approach by a diverse range of experiments on multiple datasets. Additionally we prove formally that, given enough data, FLEA protects the learner against unreliable data as long as the fraction of affected data sources is less than half.

(2 min) Adaptive gradient methods, such as \textsc{Adam}, have achieved tremendous success in machine learning. Scaling gradients by square roots of the running averages of squared past gradients, such methods are able to attain rapid training of modern deep neural networks. Nevertheless, they are observed to generalize worse than stochastic gradient descent (\textsc{SGD}) and tend to be trapped in local minima at an early stage during training. Intriguingly, we discover that substituting the gradient in the preconditioner term with the momentumized version in \textsc{Adam} can well solve the issues. The intuition is that gradient with momentum contains more accurate directional information and therefore its second moment estimation is a better choice for scaling than raw gradient's. Thereby we propose \textsc{AdaMomentum} as a new optimizer reaching the goal of training faster while generalizing better. We further develop a theory to back up the improvement in optimization and generalization and provide convergence guarantee under both convex and nonconvex settings. Extensive experiments on various models and tasks demonstrate that \textsc{AdaMomentum} exhibits comparable performance to \textsc{SGD} on vision tasks, and achieves state-of-the-art results consistently on other tasks including language processing.

(2 min) Few-shot learning aims to transfer information from one task to enable generalization on novel tasks given a few examples. This information is present both in the domain and the class labels. In this work we investigate the complementary roles of these two sources of information by combining instance-discriminative contrastive learning and supervised learning in a single framework called Supervised Momentum Contrastive learning (SUPMOCO). Our approach avoids a problem observed in supervised learning where information in images not relevant to the task is discarded, which hampers their generalization to novel tasks. We show that (self-supervised) contrastive learning and supervised learning are mutually beneficial, leading to a new state-of-the-art on the META-DATASET - a recently introduced benchmark for few-shot learning. Our method is based on a simple modification of MOCO and scales better than prior work on combining supervised and self-supervised learning. This allows us to easily combine data from multiple domains leading to further improvements.

(2 min) We introduce the framework of continuous--depth graph neural networks (GNNs). Graph neural ordinary differential equations (GDEs) are formalized as the counterpart to GNNs where the input-output relationship is determined by a continuum of GNN layers, blending discrete topological structures and differential equations. The proposed framework is shown to be compatible with various static and autoregressive GNN models. Results prove general effectiveness of GDEs: in static settings they offer computational advantages by incorporating numerical methods in their forward pass; in dynamic settings, on the other hand, they are shown to improve performance by exploiting the geometry of the underlying dynamics.

(2 min) Bayesian neural networks perform variational inference over the weights however calculation of the posterior distribution remains a challenge. Our work builds on variational inference techniques for bayesian neural networks using the original Evidence Lower Bound. In this paper, we present a stochastic bayesian neural network in which we maximize Evidence Lower Bound using a new objective function which we name as Stochastic Evidence Lower Bound. We evaluate our network on 5 publicly available UCI datasets using test RMSE and log likelihood as the evaluation metrics. We demonstrate that our work not only beats the previous state of the art algorithms but is also scalable to larger datasets.

(2 min) In this work, we train the first monolingual Lithuanian transformer model on a relatively large corpus of Lithuanian news articles and compare various output decoding algorithms for abstractive news summarization. We achieve an average ROUGE-2 score 0.163, generated summaries are coherent and look impressive at first glance. However, some of them contain misleading information that is not so easy to spot. We describe all the technical details and share our trained model and accompanying code in an online open-source repository, as well as some characteristic samples of the generated summaries.

(2 min) Genetic Programming (GP) is known to suffer from the burden of being computationally expensive by design. While, over the years, many techniques have been developed to mitigate this issue, data vectorization, in particular, is arguably still the most attractive strategy due to the parallel nature of GP. In this work, we employ a series of benchmarks meant to compare both the performance and evolution capabilities of different vectorized and iterative implementation approaches across several existing frameworks. Namely, TensorGP, a novel open-source engine written in Python, is shown to greatly benefit from the TensorFlow library to accelerate the domain evaluation phase in GP. The presented performance benchmarks demonstrate that the TensorGP engine manages to pull ahead, with relative speedups above two orders of magnitude for problems with a higher number of fitness cases. Additionally, as a consequence of being able to compute larger domains, we argue that TensorGP performance gains aid the discovery of more accurate candidate solutions.

(2 min) Temperature monitoring during the life time of heat-source components in engineering systems becomes essential to ensure the normal work and even the long working life of the heat sources. However, prior methods, which mainly use the interpolate estimation, require large amounts of temperature tensors for an accurate estimation. To solve this problem, this work develops a novel physics-informed deep surrogate models for temperature field reconstruction. First, we defines the temperature field reconstruction task of heat-source systems. Then, this work develops the deep surrogate model mapping for the proposed task. Finally, considering the physical properties of heat transfer, this work proposes four different losses and joint learns the deep surrogate model with these losses. Experimental studies have conducted over typical two-dimensional heat-source systems to demonstrate the effectiveness and efficiency of the proposed physics-informed deep surrogate models for temperature field reconstruction.

(2 min) Stochastic delays in feedback lead to unstable sequential learning using multi-armed bandits. Recently, empirical Bayesian shrinkage has been shown to improve reward estimation in bandit learning. Here, we propose a novel adaptation to shrinkage that estimates smoothed reward estimates from windowed cumulative inputs, to deal with incomplete knowledge from delayed feedback and non-stationary rewards. Using numerical simulations, we show that this adaptation retains the benefits of shrinkage, and improves the stability of reward estimation by more than 50%. Our proposal reduces variability in treatment allocations to the best arm by up to 3.8x, and improves statistical accuracy - with up to 8% improvement in true positive rates and 37% reduction in false positive rates. Together, these advantages enable control of the trade-off between speed and stability of adaptation, and facilitate human-in-the-loop sequential optimisation.

(2 min) Recently demonstrated physical-world adversarial attacks have exposed vulnerabilities in perception systems that pose severe risks for safety-critical applications such as autonomous driving. These attacks place adversarial artifacts in the physical world that indirectly cause the addition of a universal patch to inputs of a model that can fool it in a variety of contexts. Adversarial training is the most effective defense against image-dependent adversarial attacks. However, tailoring adversarial training to universal patches is computationally expensive since the optimal universal patch depends on the model weights which change during training. We propose meta adversarial training (MAT), a novel combination of adversarial training with meta-learning, which overcomes this challenge by meta-learning universal patches along with model training. MAT requires little extra computation while continuously adapting a large set of patches to the current model. MAT considerably increases robustness against universal patch attacks on image classification and traffic-light detection.

(2 min) A continuous real-time respiratory sound automated analysis system is needed in clinical practice. Previously, we established an open access lung sound database, HF_Lung_V1, and automated lung sound analysis algorithms capable of detecting inhalation, exhalation, continuous adventitious sounds (CASs) and discontinuous adventitious sounds (DASs). In this study, HF-Lung-V1 has been further expanded to HF-Lung-V2 with 1.45 times of increase in audio files. The convolutional neural network (CNN)-bidirectional gated recurrent unit (BiGRU) model was separately trained with training datasets of HF_Lung_V1 (V1_Train) and HF_Lung_V2 (V2_Train), and then were used for the performance comparisons of segment detection and event detection on both test datasets of HF_Lung_V1 (V1_Test) and HF_Lung_V2 (V2_Test). The performance of segment detection was measured by accuracy, predictive positive value (PPV), sensitivity, specificity, F1 score, receiver operating characteristic (ROC) curve and area under the curve (AUC), whereas that of event detection was evaluated with PPV, sensitivity, and F1 score. Results indicate that the model performance trained by V2_Train showed improvement on both V1_Test and V2_Test in inhalation, CASs and DASs, particularly in CASs, as well as on V1_Test in exhalation.

(2 min) At present, people usually use some methods based on convolutional neural networks (CNNs) for Electroencephalograph (EEG) decoding. However, CNNs have limitations in perceiving global dependencies, which is not adequate for common EEG paradigms with a strong overall relationship. Regarding this issue, we propose a novel EEG decoding method that mainly relies on the attention mechanism. The EEG data is firstly preprocessed and spatially filtered. And then, we apply attention transforming on the feature-channel dimension so that the model can enhance more relevant spatial features. The most crucial step is to slice the data in the time dimension for attention transforming, and finally obtain a highly distinguishable representation. At this time, global averaging pooling and a simple fully-connected layer are used to classify different categories of EEG data. Experiments on two public datasets indicate that the strategy of attention transforming effectively utilizes spatial and temporal features. And we have reached the level of the state-of-the-art in multi-classification of EEG, with fewer parameters. As far as we know, it is the first time that a detailed and complete method based on the transformer idea has been proposed in this field. It has good potential to promote the practicality of brain-computer interface (BCI). The source code can be found at: \textit{https://github.com/anranknight/EEG-Transformer}.

(2 min) Optimization problems on the generalized Stiefel manifold (and products of it) are prevalent across science and engineering. For example, in computational science they arise in the symmetric (generalized) eigenvalue problem, in nonlinear eigenvalue problems, and in electronic structures computations, to name a few problems. In statistics and machine learning, they arise, for example, in various dimensionality reduction techniques such as canonical correlation analysis. In deep learning, regularization and improved stability can be obtained by constraining some layers to have parameter matrices that belong to the Stiefel manifold. Solving problems on the generalized Stiefel manifold can be approached via the tools of Riemannian optimization. However, using the standard geometric components for the generalized Stiefel manifold has two possible shortcoming: computing some of the geometric components can be too expensive and converge can be rather slow in certain cases. Both shortcomings can be addressed using a technique called Riemannian preconditioning, which amounts to using geometric components derived using a precoditioner that defines a Riemannian metric on the constraint manifold. In this paper we develop the geometric components required to perform Riemannian optimization on the generalized Stiefel manifold equipped with a non-standard metric, and illustrate theoretically and numerically the use of those components and the effect of Riemannian preconditioning for solving optimization problems on the generalized Stiefel manifold.

(2 min) We describe a general framework -- compressive statistical learning -- for resource-efficient large-scale learning: the training collection is compressed in one pass into a low-dimensional sketch (a vector of random empirical generalized moments) that captures the information relevant to the considered learning task. A near-minimizer of the risk is computed from the sketch through the solution of a nonlinear least squares problem. We investigate sufficient sketch sizes to control the generalization error of this procedure. The framework is illustrated on compressive PCA, compressive clustering, and compressive Gaussian mixture Modeling with fixed known variance. The latter two are further developed in a companion paper.

(2 min) The spreading COVID-19 misinformation over social media already draws the attention of many researchers. According to Google Scholar, about 26000 COVID-19 related misinformation studies have been published to date. Most of these studies focusing on 1) detect and/or 2) analysing the characteristics of COVID-19 related misinformation. However, the study of the social behaviours related to misinformation is often neglected. In this paper, we introduce a fine-grained annotated misinformation tweets dataset including social behaviours annotation (e.g. comment or question to the misinformation). The dataset not only allows social behaviours analysis but also suitable for both evidence-based or non-evidence-based misinformation classification task. In addition, we introduce leave claim out validation in our experiments and demonstrate the misinformation classification performance could be significantly different when applying to real-world unseen misinformation.

(2 min) Optimization algorithms such as projected Newton's method, FISTA, mirror descent and its variants enjoy near-optimal regret bounds and convergence rates, but suffer from a computational bottleneck of computing "projections'' in potentially each iteration (e.g., $O(T^{1/2})$ regret of online mirror descent). On the other hand, conditional gradient variants solve a linear optimization in each iteration, but result in suboptimal rates (e.g., $O(T^{3/4})$ regret of online Frank-Wolfe). Motivated by this trade-off in runtime v/s convergence rates, we consider iterative projections of close-by points over widely-prevalent submodular base polytopes $B(f)$. We develop a toolkit to speed up the computation of projections using both discrete and continuous perspectives. We subsequently adapt the away-step Frank-Wolfe algorithm to use this information and enable early termination. For the special case of cardinality based submodular polytopes, we improve the runtime of computing certain Bregman projections by a factor of $\Omega(n/\log(n))$. Our theoretical results show orders of magnitude reduction in runtime in preliminary computational experiments.

(2 min) Image segmentation is a common and challenging task in autonomous driving. Availability of sufficient pixel-level annotations for the training data is a hurdle. Active learning helps learning from small amounts of data by suggesting the most promising samples for labeling. In this work, we propose a new pool-based method for active learning, which proposes promising image regions, in each acquisition step. The problem is framed in an exploration-exploitation framework by combining an embedding based on Uniform Manifold Approximation to model representativeness with entropy as uncertainty measure to model informativeness. We applied our proposed method to the challenging autonomous driving data sets CamVid and Cityscapes and performed a quantitative comparison with state-of-the-art methods. We find that our active learning method achieves better performance on CamVid compared to other methods, while on Cityscapes, the performance lift was negligible.

(2 min) With an increase in mobile and camera devices' popularity, digital content in the form of images has increased drastically. As personal life is being continuously documented in pictures, the risk of losing it to eavesdroppers is a matter of grave concern. Secondary storage is the most preferred medium for the storage of personal and other images. Our work is concerned with the security of such images. While encryption is the best way to ensure image security, full encryption and decryption is a computationally-intensive process. Moreover, as cameras are getting better every day, image quality, and thus, the pixel density has increased considerably. The increased pixel density makes encryption and decryption more expensive. We, therefore, delve into selective encryption and selective blurring based on the region of interest. Instead of encrypting or blurring the entire photograph, we only encode selected regions of the image. We present a comparative analysis of the partial and full encryption of the photos. This kind of encoding will help us lower the encryption overhead without compromising security. The applications utilizing this technique will become more usable due to the reduction in the decryption time. Additionally, blurred images being more readable than encrypted ones, allowed us to define the level of security. We leverage the machine learning algorithms like Mask-RCNN (Region-based convolutional neural network) and YOLO (You Only Look Once) to select the region of interest. These algorithms have set new benchmarks for object recognition. We develop an end to end system to demonstrate our idea of selective encryption.

(2 min) We consider the optimal approximate posterior over the top-layer weights in a Bayesian neural network for regression, and show that it exhibits strong dependencies on the lower-layer weights. We adapt this result to develop a correlated approximate posterior over the weights at all layers in a Bayesian neural network. We extend this approach to deep Gaussian processes, unifying inference in the two model classes. Our approximate posterior uses learned "global" inducing points, which are defined only at the input layer and propagated through the network to obtain inducing inputs at subsequent layers. By contrast, standard, "local", inducing point methods from the deep Gaussian process literature optimise a separate set of inducing inputs at every layer, and thus do not model correlations across layers. Our method gives state-of-the-art performance for a variational Bayesian method, without data augmentation or tempering, on CIFAR-10 of 86.7%, which is comparable to SGMCMC without tempering but with data augmentation (88% in Wenzel et al. 2020).

(2 min) We prove that every concept class with finite Littlestone dimension can be learned by an (approximate) differentially-private algorithm. This answers an open question of Alon et al. (STOC 2019) who proved the converse statement (this question was also asked by Neel et al.~(FOCS 2019)). Together these two results yield an equivalence between online learnability and private PAC learnability. We introduce a new notion of algorithmic stability called "global stability" which is essential to our proof and may be of independent interest. We also discuss an application of our results to boosting the privacy and accuracy parameters of differentially-private learners.

(2 min) As financial services (FS) companies have experienced drastic technology driven changes, the availability of new data streams provides the opportunity for more comprehensive customer understanding. We propose Dynamic Customer Embeddings (DCE), a framework that leverages customers' digital activity and a wide range of financial context to learn dense representations of customers in the FS industry. Our method examines customer actions and pageviews within a mobile or web digital session, the sequencing of the sessions themselves, and snapshots of common financial features across our organization at the time of login. We test our customer embeddings using real world data in three prediction problems: 1) the intent of a customer in their next digital session, 2) the probability of a customer calling the call centers after a session, and 3) the probability of a digital session to be fraudulent. DCE showed performance lift in all three downstream problems.

(2 min) Discovering the partial differential equations underlying a spatio-temporal datasets from very limited observations is of paramount interest in many scientific fields. However, it remains an open question to know when model discovery algorithms based on sparse regression can actually recover the underlying physical processes. We trace back the poor of performance of Lasso based model discovery algorithms to its potential variable selection inconsistency: meaning that even if the true model is present in the library, it might not be selected. By first revisiting the irrepresentability condition (IRC) of the Lasso, we gain some insights of when this might occur. We then show that the adaptive Lasso will have more chances of verifying the IRC than the Lasso and propose to integrate it within a deep learning model discovery framework with stability selection and error control. Experimental results show we can recover several nonlinear and chaotic canonical PDEs with a single set of hyperparameters from a very limited number of samples at high noise levels.

(2 min) Graph neural networks (GNNs) are designed for semi-supervised node classification on graphs where only a small subset of nodes have class labels. However, under extreme cases when very few labels are available (e.g., 1 labeled node per class), GNNs suffer from severe result quality degradation. Several existing studies make an initial effort to ease this situation, but are still far from satisfactory. In this paper, on few-labeled graph data, we propose an effective framework ABN that is readily applicable to both shallow and deep GNN architectures and significantly boosts classification accuracy. In particular, on a benchmark dataset Cora with only 1 labeled node per class, while the classic graph convolutional network (GCN) only has 44.6% accuracy, an immediate instantiation of ABN over GCN achieves 62.5% accuracy; when applied to a deep architecture DAGNN, ABN improves accuracy from 59.8% to 66.4%, which is state of the art. ABN obtains superior performance through three main algorithmic designs. First, it selects high-quality unlabeled nodes via an adaptive pseudo labeling technique, so as to adaptively enhance the training process of GNNs. Second, ABN balances the labels of the selected nodes on real-world skewed graph data by pseudo label balancing. Finally, a negative sampling regularizer is designed for ABN to further utilize the unlabeled nodes. The effectiveness of the three techniques in ABN is well-validated by both theoretical and empirical analysis. Extensive experiments, comparing 12 existing approaches on 4 benchmark datasets, demonstrate that ABN achieves state-of-the-art performance.

(2 min) In class-incremental learning, an agent with limited resources needs to learn a sequence of classification tasks, forming an ever growing classification problem, with the constraint of not being able to access data from previous tasks. The main difference with task-incremental learning, where a task-ID is available at inference time, is that the learner also needs to perform cross-task discrimination, i.e. distinguish between classes that have not been seen together. Approaches to tackle this problem are numerous and mostly make use of an external memory (buffer) of non-negligible size. In this paper, we ablate the learning of cross-task features and study its influence on the performance of basic replay strategies used for class-IL. We also define a new forgetting measure for class-incremental learning, and see that forgetting is not the principal cause of low performance. Our experimental results show that future algorithms for class-incremental learning should not only prevent forgetting, but also aim to improve the quality of the cross-task features. This is especially important when the number of classes per task is small.

(2 min) Modeling dynamical systems plays a crucial role in capturing and understanding complex physical phenomena. When physical models are not sufficiently accurate or hardly describable by analytical formulas, one can use generic function approximators such as neural networks to capture the system dynamics directly from sensor measurements. As for now, current methods to learn the parameters of these neural networks are highly sensitive to the inherent instability of most dynamical systems of interest, which in turn prevents the study of very long sequences. In this work, we introduce a generic and scalable method based on multiple shooting to learn latent representations of indirectly observed dynamical systems. We achieve state-of-the-art performances on systems observed directly from raw images. Further, we demonstrate that our method is robust to noisy measurements and can handle complex dynamical systems, such as chaotic ones.

(2 min) In this paper, we consider the framework of privacy amplification via iteration, which is originally proposed by Feldman et al. and subsequently simplified by Asoodeh et al. in their analysis via the contraction coefficient. This line of work focuses on the study of the privacy guarantees obtained by the projected noisy stochastic gradient descent (PNSGD) algorithm with hidden intermediate updates. A limitation in the existing literature is that only the early stopped PNSGD has been studied, while no result has been proved on the more widely-used PNSGD applied on a shuffled dataset. Moreover, no scheme has been yet proposed regarding how to decrease the injected noise when new data are received in an online fashion. In this work, we first prove a privacy guarantee for shuffled PNSGD, which is investigated asymptotically when the noise is fixed for each sample size $n$ but reduced at a predetermined rate when $n$ increases, in order to achieve the convergence of privacy loss. We then analyze the online setting and provide a faster decaying scheme for the magnitude of the injected noise that also guarantees the convergence of privacy loss.

(2 min) We study the problem of estimating a rank-one matrix from Gaussian observations where different blocks of the matrix are observed under different noise levels. This problem is motivated by applications in clustering and community detection where latent variables can be partitioned into a fixed number of known groups (e.g., users and items) and the blocks of the matrix correspond to different types of pairwise interactions (e.g., user-user, user-item, or item-item interactions). In the setting where the number of blocks is fixed while the number of variables tends to infinity, we prove asymptotically exact formulas for the minimum mean-squared error in estimating both the matrix and the latent variables. These formulas describe the weak recovery thresholds for the problem and reveal invariance properties with respect to certain scalings of the noise variance. We also derive an approximate message passing algorithm and a gradient descent algorithm and show empirically that these algorithms achieve the information-theoretic limits in certain regimes.

(2 min) We study deep reinforcement learning (RL) algorithms with delayed rewards. In many real-world tasks, instant rewards are often not readily accessible or even defined immediately after the agent performs actions. In this work, we first formally define the environment with delayed rewards and discuss the challenges raised due to the non-Markovian nature of such environments. Then, we introduce a general off-policy RL framework with a new Q-function formulation that can handle the delayed rewards with theoretical convergence guarantees. For practical tasks with high dimensional state spaces, we further introduce the HC-decomposition rule of the Q-function in our framework which naturally leads to an approximation scheme that helps boost the training efficiency and stability. We finally conduct extensive experiments to demonstrate the superior performance of our algorithms over the existing work and their variants.

(2 min) Organ-at-risk contouring is still a bottleneck in radiotherapy, with many deep learning methods falling short of promised results when evaluated on clinical data. We investigate the accuracy and time-savings resulting from the use of an interactive-machine-learning method for an organ-at-risk contouring task. We compare the method to the Eclipse contouring software and find strong agreement with manual delineations, with a dice score of 0.95. The annotations created using corrective-annotation also take less time to create as more images are annotated, resulting in substantial time savings compared to manual methods, with hearts that take 2 minutes and 2 seconds to delineate on average, after 923 images have been delineated, compared to 7 minutes and 1 seconds when delineating manually. Our experiment demonstrates that interactive-machine-learning with corrective-annotation provides a fast and accessible way for non computer-scientists to train deep-learning models to segment their own structures of interest as part of routine clinical workflows. Source code is available at \href{https://github.com/Abe404/RootPainter3D}{this HTTPS URL}.

(2 min) Federated learning is an emerging machine learning paradigm that enables multiple devices to train models locally and formulate a global model, without sharing the clients' local data. A federated learning system can be viewed as a large-scale distributed system, involving different components and stakeholders with diverse requirements and constraints. Hence, developing a federated learning system requires both software system design thinking and machine learning knowledge. Although much effort has been put into federated learning from the machine learning perspectives, our previous systematic literature review on the area shows that there is a distinct lack of considerations for software architecture design for federated learning. In this paper, we propose FLRA, a reference architecture for federated learning systems, which provides a template design for federated learning-based solutions. The proposed FLRA reference architecture is based on an extensive review of existing patterns of federated learning systems found in the literature and existing industrial implementation. The FLRA reference architecture consists of a pool of architectural patterns that could address the frequently recurring design problems in federated learning architectures. The FLRA reference architecture can serve as a design guideline to assist architects and developers with practical solutions for their problems, which can be further customised.

(2 min) Approximate Bayesian inference for neural networks is considered a robust alternative to standard training, often providing good performance on out-of-distribution data. However, Bayesian neural networks (BNNs) with high-fidelity approximate inference via full-batch Hamiltonian Monte Carlo achieve poor generalization under covariate shift, even underperforming classical estimation. We explain this surprising result, showing how a Bayesian model average can in fact be problematic under covariate shift, particularly in cases where linear dependencies in the input features cause a lack of posterior contraction. We additionally show why the same issue does not affect many approximate inference procedures, or classical maximum a-posteriori (MAP) training. Finally, we propose novel priors that improve the robustness of BNNs to many sources of covariate shift.

(2 min) Automatic code synthesis from natural language descriptions is a challenging task. We witness massive progress in developing code generation systems for domain-specific languages (DSLs) employing sequence-to-sequence deep learning techniques in the recent past. In this paper, we specifically experiment with \textsc{AlgoLisp} DSL-based generative models and showcase the existence of significant dataset bias through different classes of adversarial examples. We also experiment with two variants of Transformer-based models that outperform all existing \textsc{AlgoLisp} DSL-based code generation baselines. Consistent with the current state-of-the-art systems, our proposed models, too, achieve poor performance under adversarial settings. Therefore, we propose several dataset augmentation techniques to reduce bias and showcase their efficacy using robust experimentation.

(2 min) The author of this work proposes an overview of the recent semi-supervised learning approaches and related works. Despite the remarkable success of neural networks in various applications, there exist few formidable constraints including the need for a large amount of labeled data. Therefore, semi-supervised learning, which is a learning scheme in which the scarce labels and a larger amount of unlabeled data are utilized to train models (e.g., deep neural networks) is getting more important. Based on the key assumptions of semi-supervised learning, which are the manifold assumption, cluster assumption, and continuity assumption, the work reviews the recent semi-supervised learning approaches. In particular, the methods in regard to using deep neural networks in a semi-supervised learning setting are primarily discussed. In addition, the existing works are first classified based on the underlying idea and explained, and then the holistic approaches that unify the aforementioned ideas are detailed.

(2 min) A variety of wireless channel estimation methods, e.g., MUSIC and ESPRIT, rely on prior knowledge of the model order. Therefore, it is important to correctly estimate the number of multipath components (MPCs) which compose such channels. However, environments with many scatterers may generate MPCs which are closely spaced. This clustering of MPCs in addition to noise makes the model order selection task difficult in practice to currently known algorithms. In this paper, we exploit the multidimensional characteristics of MIMO orthogonal frequency division multiplexing (OFDM) systems and propose a machine learning (ML) method capable of determining the number of MPCs with a higher accuracy than state of the art methods in almost coherent scenarios. Moreover, our results show that our proposed ML method has an enhanced reliability.

(2 min) Recently, Graph Neural Network (GNN) has achieved remarkable success in various real-world problems on graph data. However in most industries, data exists in the form of isolated islands and the data privacy and security is also an important issue. In this paper, we propose FedVGCN, a federated GCN learning paradigm for privacy-preserving node classification task under data vertically partitioned setting, which can be generalized to existing GCN models. Specifically, we split the computation graph data into two parts. For each iteration of the training process, the two parties transfer intermediate results to each other under homomorphic encryption. We conduct experiments on benchmark data and the results demonstrate the effectiveness of FedVGCN in the case of GraphSage.

(2 min) With leveraging the weight-sharing and continuous relaxation to enable gradient-descent to alternately optimize the supernet weights and the architecture parameters through a bi-level optimization paradigm, \textit{Differentiable ARchiTecture Search} (DARTS) has become the mainstream method in Neural Architecture Search (NAS) due to its simplicity and efficiency. However, more recent works found that the performance of the searched architecture barely increases with the optimization proceeding in DARTS. In addition, several concurrent works show that the NAS could find more competitive architectures without labels. The above observations reveal that the supervision signal in DARTS may be a poor indicator for architecture optimization, inspiring a foundational question: instead of using the supervision signal to perform bi-level optimization, \textit{can we find high-quality architectures \textbf{without any training nor labels}}? We provide an affirmative answer by customizing the NAS as a network pruning at initialization problem. By leveraging recent techniques on the network pruning at initialization, we designed a FreeFlow proxy to score the importance of candidate operations in NAS without any training nor labels, and proposed a novel framework called \textit{training and label free neural architecture search} (\textbf{FreeNAS}) accordingly. We show that, without any training nor labels, FreeNAS with the proposed FreeFlow proxy can outperform most NAS baselines. More importantly, our framework is extremely efficient, which completes the architecture search within only \textbf{3.6s} and \textbf{79s} on a single GPU for the NAS-Bench-201 and DARTS search space, respectively. We hope our work inspires more attempts in solving NAS from the perspective of pruning at initialization.

(2 min) Logical relations widely exist in human activities. Human use them for making judgement and decision according to various conditions, which are embodied in the form of \emph{if-then} rules. As an important kind of cognitive intelligence, it is prerequisite of representing and storing logical relations rightly into computer systems so as to make automatic judgement and decision, especially for high-risk domains like medical diagnosis. However, current numeric ANN (Artificial Neural Network) models are good at perceptual intelligence such as image recognition while they are not good at cognitive intelligence such as logical representation, blocking the further application of ANN. To solve it, researchers have tried to design logical ANN models to represent and store logical relations. Although there are some advances in this research area, recent works still have disadvantages because the structures of these logical ANN models still don't map more directly with logical relations which will cause the corresponding logical relations cannot be read out from their network structures. Therefore, in order to represent logical relations more clearly by the neural network structure and to read out logical relations from it, this paper proposes a novel logical ANN model by designing the new logical neurons and links in demand of logical representation. Compared with the recent works on logical ANN models, this logical ANN model has more clear corresponding with logical relations using the more direct mapping method herein, thus logical relations can be read out following the connection patterns of the network structure. Additionally, less neurons are used.

(2 min) Deep ensembles have recently gained popularity in the deep learning community for their conceptual simplicity and efficiency. However, maintaining functional diversity between ensemble members that are independently trained with gradient descent is challenging. This can lead to pathologies when adding more ensemble members, such as a saturation of the ensemble performance, which converges to the performance of a single model. Moreover, this does not only affect the quality of its predictions, but even more so the uncertainty estimates of the ensemble, and thus its performance on out-of-distribution data. We hypothesize that this limitation can be overcome by discouraging different ensemble members from collapsing to the same function. To this end, we introduce a kernelized repulsive term in the update rule of the deep ensembles. We show that this simple modification not only enforces and maintains diversity among the members but, even more importantly, transforms the maximum a posteriori inference into proper Bayesian inference. Namely, we show that the training dynamics of our proposed repulsive ensembles follow a Wasserstein gradient flow of the KL divergence with the true posterior. We study repulsive terms in weight and function space and empirically compare their performance to standard ensembles and Bayesian baselines on synthetic and real-world prediction tasks.

(2 min) Kernel segmentation aims at partitioning a data sequence into several non-overlapping segments that may have nonlinear and complex structures. In general, it is formulated as a discrete optimization problem with combinatorial constraints. A popular algorithm for optimally solving this problem is dynamic programming (DP), which has quadratic computation and memory requirements. Given that sequences in practice are too long, this algorithm is not a practical approach. Although many heuristic algorithms have been proposed to approximate the optimal segmentation, they have no guarantee on the quality of their solutions. In this paper, we take a differentiable approach to alleviate the aforementioned issues. First, we introduce a novel sigmoid-based regularization to smoothly approximate the combinatorial constraints. Combining it with objective of the balanced kernel clustering, we formulate a differentiable model termed Kernel clustering with sigmoid-based regularization (KCSR), where the gradient-based algorithm can be exploited to obtain the optimal segmentation. Second, we develop a stochastic variant of the proposed model. By using the stochastic gradient descent algorithm, which has much lower time and space complexities, for optimization, the second model can perform segmentation on overlong data sequences. Finally, for simultaneously segmenting multiple data sequences, we slightly modify the sigmoid-based regularization to further introduce an extended variant of the proposed model. Through extensive experiments on various types of data sequences performances of our models are evaluated and compared with those of the existing methods. The experimental results validate advantages of the proposed models. Our Matlab source code is available on github.

(2 min) Model checking has been developed for verifying the behaviour of systems with stochastic and non-deterministic behavior. It is used to provide guarantees about such systems. While most model checking methods focus on propositional models, various probabilistic planning and reinforcement frameworks deal with relational domains, for instance, STRIPS planning and relational Markov Decision Processes. Using propositional model checking in relational settings requires one to ground the model, which leads to the well known state explosion problem and intractability. We present pCTL-REBEL, a lifted model checking approach for verifying pCTL properties on relational MDPs. It extends REBEL, the relational Bellman update operator, which is a lifted value iteration approach for model-based relational reinforcement learning, toward relational model-checking. PCTL-REBEL is lifted, which means that rather than grounding, the model exploits symmetries and reasons at an abstract relational level. Theoretically, we show that the pCTL model checking approach is decidable for relational MDPs even for possibly infinite domains provided that the states have a bounded size. Practically, we contribute algorithms and an implementation of lifted relational model checking, and we show that the lifted approach improves the scalability of the model checking approach.

(2 min) In this chapter, we will give comprehensive examples of applying RL in optimizing the physical layer of wireless communications by defining different class of problems and the possible solutions to handle them. In Section 9.2, we present all the basic theory needed to address a RL problem, i.e. Markov decision process (MDP), Partially observable Markov decision process (POMDP), but also two very important and widely used algorithms for RL, i.e. the Q-learning and SARSA algorithms. We also introduce the deep reinforcement learning (DRL) paradigm and the section ends with an introduction to the multi-armed bandits (MAB) framework. Section 9.3 focuses on some toy examples to illustrate how the basic concepts of RL are employed in communication systems. We present applications extracted from literature with simplified system models using similar notation as in Section 9.2 of this Chapter. In Section 9.3, we also focus on modeling RL problems, i.e. how action and state spaces and rewards are chosen. The Chapter is concluded in Section 9.4 with a prospective thought on RL trends and it ends with a review of a broader state of the art in Section 9.5.

(2 min) We present a number of low-resource approaches to the tasks of the Zero Resource Speech Challenge 2021. We build on the unsupervised representations of speech proposed by the organizers as a baseline, derived from CPC and clustered with the k-means algorithm. We demonstrate that simple methods of refining those representations can narrow the gap, or even improve upon the solutions which use a high computational budget. The results lead to the conclusion that the CPC-derived representations are still too noisy for training language models, but stable enough for simpler forms of pattern matching and retrieval.

(2 min) In high energy physics (HEP), jets are collections of correlated particles produced ubiquitously in particle collisions such as those at the CERN Large Hadron Collider (LHC). Machine-learning-based generative models, such as generative adversarial networks (GANs), have the potential to significantly accelerate LHC jet simulations. However, despite jets having a natural representation as a set of particles in momentum-space, a.k.a. a particle cloud, to our knowledge there exist no generative models applied to such a dataset. We introduce a new particle cloud dataset (JetNet), and, due to similarities between particle and point clouds, apply to it existing point cloud GANs. Results are evaluated using (1) the 1-Wasserstein distance between high- and low-level feature distributions, (2) a newly developed Fr\'{e}chet ParticleNet Distance, and (3) the coverage and (4) minimum matching distance metrics. Existing GANs are found to be inadequate for physics applications, hence we develop a new message passing GAN (MPGAN), which outperforms existing point cloud GANs on virtually every metric and shows promise for use in HEP. We propose JetNet as a novel point-cloud-style dataset for the machine learning community to experiment with, and set MPGAN as a benchmark to improve upon for future generative models.

(2 min) Treatment effect estimation from observational data is a fundamental problem in causal inference. There are two very different schools of thought that have tackled this problem. On the one hand, the Pearlian framework commonly assumes structural knowledge (provided by an expert) in the form of Directed Acyclic Graphs (DAGs) and provides graphical criteria such as the back-door criterion to identify the valid adjustment sets. On the other hand, the potential outcomes (PO) framework commonly assumes that all the observed features satisfy ignorability (i.e., no hidden confounding), which in general is untestable. In this work, we take steps to bridge these two frameworks. We show that even if we know only one parent of the treatment variable (provided by an expert), then quite remarkably it suffices to test a broad class of (but not all) back-door criteria. Importantly, we also cover the non-trivial case where the entire set of observed features is not ignorable (generalizing the PO framework) without requiring all the parents of the treatment variable to be observed. Our key technical idea involves a more general result -- Given a synthetic sub-sampling (or environment) variable that is a function of the parent variable, we show that an invariance test involving this sub-sampling variable is equivalent to testing a broad class of back-door criteria. We demonstrate our approach on synthetic data as well as real causal effect estimation benchmarks.

(2 min) Social media platforms bring together content creators and content consumers through recommender systems like newsfeed. The focus of such recommender systems has thus far been primarily on modeling the content consumer preferences and optimizing for their experience. However, it is equally critical to nurture content creation by prioritizing the creators' interests, as quality content forms the seed for sustainable engagement and conversations, bringing in new consumers while retaining existing ones. In this work, we propose a modeling approach to predict how feedback from content consumers incentivizes creators. We then leverage this model to optimize the newsfeed experience for content creators by reshaping the feedback distribution, leading to a more active content ecosystem. Practically, we discuss how we balance the user experience for both consumers and creators, and how we carry out online A/B tests with strong network effects. We present a deployed use case on the LinkedIn newsfeed, where we used this approach to improve content creation significantly without compromising the consumers' experience.

(2 min) Moving Object Detection (MOD) is a crucial task for the Autonomous Driving pipeline. MOD is usually handled via 2-stream convolutional architectures that incorporates both appearance and motion cues, without considering the inter-relations between the spatial or motion features. In this paper, we tackle this problem through multi-head attention mechanisms, both across the spatial and motion streams. We propose MODETR; a Moving Object DEtection TRansformer network, comprised of multi-stream transformer encoders for both spatial and motion modalities, and an object transformer decoder that produces the moving objects bounding boxes using set predictions. The whole architecture is trained end-to-end using bi-partite loss. Several methods of incorporating motion cues with the Transformer model are explored, including two-stream RGB and Optical Flow (OF) methods, and multi-stream architectures that take advantage of sequence information. To incorporate the temporal information, we propose a new Temporal Positional Encoding (TPE) approach to extend the Spatial Positional Encoding(SPE) in DETR. We explore two architectural choices for that, balancing between speed and time. To evaluate the our network, we perform the MOD task on the KITTI MOD [6] data set. Results show significant 5% mAP of the Transformer network for MOD over the state-of-the art methods. Moreover, the proposed TPE encoding provides 10% mAP improvement over the SPE baseline.

(2 min) In this paper, we present a physics-constrained deep neural network (PCDNN) method for parameter estimation in the zero-dimensional (0D) model of the vanadium redox flow battery (VRFB). In this approach, we use deep neural networks (DNNs) to approximate the model parameters as functions of the operating conditions. This method allows the integration of the VRFB computational models as the physical constraints in the parameter learning process, leading to enhanced accuracy of parameter estimation and cell voltage prediction. Using an experimental dataset, we demonstrate that the PCDNN method can estimate model parameters for a range of operating conditions and improve the 0D model prediction of voltage compared to the 0D model prediction with constant operation-condition-independent parameters estimated with traditional inverse methods. We also demonstrate that the PCDNN approach has an improved generalization ability for estimating parameter values for operating conditions not used in the DNN training.

(2 min) A photoplethysmography (PPG) is an uncomplicated and inexpensive optical technique widely used in the healthcare domain to extract valuable health-related information, e.g., heart rate variability, blood pressure, and respiration rate. PPG signals can easily be collected continuously and remotely using portable wearable devices. However, these measuring devices are vulnerable to motion artifacts caused by daily life activities. The most common ways to eliminate motion artifacts use extra accelerometer sensors, which suffer from two limitations: i) high power consumption and ii) the need to integrate an accelerometer sensor in a wearable device (which is not required in certain wearables). This paper proposes a low-power non-accelerometer-based PPG motion artifacts removal method outperforming the accuracy of the existing methods. We use Cycle Generative Adversarial Network to reconstruct clean PPG signals from noisy PPG signals. Our novel machine-learning-based technique achieves 9.5 times improvement in motion artifact removal compared to the state-of-the-art without using extra sensors such as an accelerometer.

(2 min) The study of provable adversarial robustness for deep neural network (DNN) models has mainly focused on static supervised learning tasks such as image classification. However, DNNs have been used extensively in real-world adaptive tasks such as reinforcement learning (RL), making RL systems vulnerable to adversarial attacks. The key challenge in adversarial RL is that the attacker can adapt itself to the defense strategy used by the agent in previous time-steps to strengthen its attack in future steps. In this work, we study the provable robustness of RL against norm-bounded adversarial perturbations of the inputs. We focus on smoothing-based provable defenses and propose policy smoothing where the agent adds a Gaussian noise to its observation at each time-step before applying the policy network to make itself less sensitive to adversarial perturbations of its inputs. Our main theoretical contribution is to prove an adaptive version of the Neyman-Pearson Lemma where the adversarial perturbation at a particular time can be a stochastic function of current and previous observations and states as well as previously observed actions. Using this lemma, we adapt the robustness certificates produced by randomized smoothing in the static setting of image classification to the dynamic setting of RL. We generate certificates that guarantee that the total reward obtained by the smoothed policy will not fall below a certain threshold under a norm-bounded adversarial perturbation of the input. We show that our certificates are tight by constructing a worst-case setting that achieves the bounds derived in our analysis. In our experiments, we show that this method can yield meaningful certificates in complex environments demonstrating its effectiveness against adversarial attacks.

(2 min) The optimal design of neural networks is a critical problem in many applications. Here, we investigate how dynamical systems with polynomial nonlinearities can inform the design of neural systems that seek to emulate them. We propose a Learnability metric and its associated features to quantify the near-equilibrium behavior of learning dynamics. Equating the Learnability of neural systems with equivalent parameter estimation metric of the reference system establishes bounds on network structure. In this way, norms from theory provide a good first guess for neural structure, which may then further adapt with data. The proposed approach neither requires training nor training data. It reveals exact sizing for a class of neural networks with multiplicative nodes that mimic continuous- or discrete-time polynomial dynamics. It also provides relatively tight lower size bounds for classical feed-forward networks that is consistent with simulated assessments.

(2 min) Most modern neural networks for classification fail to take into account the concept of the unknown. Trained neural networks are usually tested in an unrealistic scenario with only examples from a closed set of known classes. In an attempt to develop a more realistic model, the concept of working in an open set environment has been introduced. This in turn leads to the concept of incremental learning where a model with its own architecture and initial trained set of data can identify unknown classes during the testing phase and autonomously update itself if evidence of a new class is detected. Some problems that arise in incremental learning are inefficient use of resources to retrain the classifier repeatedly and the decrease of classification accuracy as multiple classes are added over time. This process of instantiating new classes is repeated as many times as necessary, accruing errors. To address these problems, this paper proposes the Classification Confidence Threshold approach to prime neural networks for incremental learning to keep accuracies high by limiting forgetting. A lean method is also used to reduce resources used in the retraining of the neural network. The proposed method is based on the idea that a network is able to incrementally learn a new class even when exposed to a limited number samples associated with the new class. This method can be applied to most existing neural networks with minimal changes to network architecture.

(2 min) Recently deep reinforcement learning has achieved tremendous success in wide ranges of applications. However, it notoriously lacks data-efficiency and interpretability. Data-efficiency is important as interacting with the environment is expensive. Further, interpretability can increase the transparency of the black-box-style deep RL models and hence gain trust from the users. In this work, we propose a new hierarchical framework via symbolic RL, leveraging a symbolic transition model to improve the data-efficiency and introduce the interpretability for learned policy. This framework consists of a high-level agent, a subtask solver and a symbolic transition model. Without assuming any prior knowledge on the state transition, we adopt inductive logic programming (ILP) to learn the rules of symbolic state transitions, introducing interpretability and making the learned behavior understandable to users. In empirical experiments, we confirmed that the proposed framework offers approximately between 30\% to 40\% more data efficiency over previous methods.

(2 min) State-of-the-art summarization systems are trained and evaluated on massive datasets scraped from the web. Despite their prevalence, we know very little about the underlying characteristics (data noise, summarization complexity, etc.) of these datasets, and how these affect system performance and the reliability of automatic metrics like ROUGE. In this study, we manually analyze 600 samples from three popular summarization datasets. Our study is driven by a six-class typology which captures different noise types (missing facts, entities) and degrees of summarization difficulty (extractive, abstractive). We follow with a thorough analysis of 27 state-of-the-art summarization models and 5 popular metrics, and report our key insights: (1) Datasets have distinct data quality and complexity distributions, which can be traced back to their collection process. (2) The performance of models and reliability of metrics is dependent on sample complexity. (3) Faithful summaries often receive low scores because of the poor diversity of references. We release the code, annotated data and model outputs.

(2 min) Routing methods in capsule networks often learn a hierarchical relationship for capsules in successive layers, but the intra-relation between capsules in the same layer is less studied, while this intra-relation is a key factor for the semantic understanding in text data. Therefore, in this paper, we introduce a new capsule network with graph routing to learn both relationships, where capsules in each layer are treated as the nodes of a graph. We investigate strategies to yield adjacency and degree matrix with three different distances from a layer of capsules, and propose the graph routing mechanism between those capsules. We validate our approach on five text classification datasets, and our findings suggest that the approach combining bottom-up routing and top-down attention performs the best. Such an approach demonstrates generalization capability across datasets. Compared to the state-of-the-art routing methods, the improvements in accuracy in the five datasets we used were 0.82, 0.39, 0.07, 1.01, and 0.02, respectively.

(2 min) Recent progress on deep learning relies heavily on the quality and efficiency of training algorithms. In this paper, we develop a fast training method motivated by the nonlinear Conjugate Gradient (CG) framework. We propose the Conjugate Gradient with Quadratic line-search (CGQ) method. On the one hand, a quadratic line-search determines the step size according to current loss landscape. On the other hand, the momentum factor is dynamically updated in computing the conjugate gradient parameter (like Polak-Ribiere). Theoretical results to ensure the convergence of our method in strong convex settings is developed. And experiments in image classification datasets show that our method yields faster convergence than other local solvers and has better generalization capability (test set accuracy). One major advantage of the paper method is that tedious hand tuning of hyperparameters like the learning rate and momentum is avoided.

(2 min) We study mean-field variational Bayesian inference using the TAP approach, for Z2-synchronization as a prototypical example of a high-dimensional Bayesian model. We show that for any signal strength $\lambda > 1$ (the weak-recovery threshold), there exists a unique local minimizer of the TAP free energy functional near the mean of the Bayes posterior law. Furthermore, the TAP free energy in a local neighborhood of this minimizer is strongly convex. Consequently, a natural-gradient/mirror-descent algorithm achieves linear convergence to this minimizer from a local initialization, which may be obtained by a finite number of iterates of Approximate Message Passing (AMP). This provides a rigorous foundation for variational inference in high dimensions via minimization of the TAP free energy. We also analyze the finite-sample convergence of AMP, showing that AMP is asymptotically stable at the TAP minimizer for any $\lambda > 1$, and is linearly convergent to this minimizer from a spectral initialization for sufficiently large $\lambda$. Such a guarantee is stronger than results obtainable by state evolution analyses, which only describe a fixed number of AMP iterations in the infinite-sample limit. Our proofs combine the Kac-Rice formula and Sudakov-Fernique Gaussian comparison inequality to analyze the complexity of critical points that satisfy strong convexity and stability conditions within their local neighborhoods.

(2 min) Several recent studies have shown that Deep Neural Network (DNN)-based classifiers are vulnerable against model extraction attacks. In model extraction attacks, an adversary exploits the target classifier to create a surrogate classifier imitating the target classifier with respect to some criteria. In this paper, we investigate the hardness degree of samples and demonstrate that the hardness degree histogram of model extraction attacks samples is distinguishable from the hardness degree histogram of normal samples. Normal samples come from the target classifier's training data distribution. As the training process of DNN-based classifiers is done in several epochs, we can consider this process as a sequence of subclassifiers so that each subclassifier is created at the end of an epoch. We use the sequence of subclassifiers to calculate the hardness degree of samples. We investigate the relation between hardness degree of samples and the trust in the classifier outputs. We propose Hardness-Oriented Detection Approach (HODA) to detect the sample sequences of model extraction attacks. The results demonstrate that HODA can detect the sample sequences of model extraction attacks with a high success rate by only watching 100 attack samples. We also investigate the hardness degree of adversarial examples and indicate that the hardness degree histogram of adversarial examples is distinct from the hardness degree histogram of normal samples.

(2 min) Place Recognition is a crucial capability for mobile robot localization and navigation. Image-based or Visual Place Recognition (VPR) is a challenging problem as scene appearance and camera viewpoint can change significantly when places are revisited. Recent VPR methods based on ``sequential representations'' have shown promising results as compared to traditional sequence score aggregation or single image based techniques. In parallel to these endeavors, 3D point clouds based place recognition is also being explored following the advances in deep learning based point cloud processing. However, a key question remains: is an explicit 3D structure based place representation always superior to an implicit ``spatial'' representation based on sequence of RGB images which can inherently learn scene structure. In this extended abstract, we attempt to compare these two types of methods by considering a similar ``metric span'' to represent places. We compare a 3D point cloud based method (PointNetVLAD) with image sequence based methods (SeqNet and others) and showcase that image sequence based techniques approach, and can even surpass, the performance achieved by point cloud based methods for a given metric span. These performance variations can be attributed to differences in data richness of input sensors as well as data accumulation strategies for a mobile robot. While a perfect apple-to-apple comparison may not be feasible for these two different modalities, the presented comparison takes a step in the direction of answering deeper questions regarding spatial representations, relevant to several applications like Autonomous Driving and Augmented/Virtual Reality. Source code available publicly https://github.com/oravus/seqNet.

(2 min) Top-down attention allows neural networks, both artificial and biological, to focus on the information most relevant for a given task. This is known to enhance performance in visual perception. But it remains unclear how attention brings about its perceptual boost, especially when it comes to naturalistic settings like recognising an object in an everyday scene. What aspects of a visual task does attention help to deal with? We aim to answer this with a computational experiment based on a general framework called task-oriented ablation design. First we define a broad range of visual tasks and identify six factors that underlie task variability. Then on each task we compare the performance of two neural networks, one with top-down attention and one without. These comparisons reveal the task-dependence of attention's perceptual boost, giving a clearer idea of the role attention plays. Whereas many existing cognitive accounts link attention to stimulus-level variables, such as visual clutter and object scale, we find greater explanatory power in system-level variables that capture the interaction between the model, the distribution of training data and the task format. This finding suggests a shift in how attention is studied could be fruitful. We make publicly available our code and results, along with statistics relevant to ImageNet-based experiments beyond this one. Our contribution serves to support the development of more human-like vision models and the design of more informative machine-learning experiments.

(2 min) Electronic health records represent a holistic overview of patients' trajectories. Their increasing availability has fueled new hopes to leverage them and develop accurate risk prediction models for a wide range of diseases. Given the complex interrelationships of medical records and patient outcomes, deep learning models have shown clear merits in achieving this goal. However, a key limitation of these models remains their capacity in processing long sequences. Capturing the whole history of medical encounters is expected to lead to more accurate predictions, but the inclusion of records collected for decades and from multiple resources can inevitably exceed the receptive field of the existing deep learning architectures. This can result in missing crucial, long-term dependencies. To address this gap, we present Hi-BEHRT, a hierarchical Transformer-based model that can significantly expand the receptive field of Transformers and extract associations from much longer sequences. Using a multimodal large-scale linked longitudinal electronic health records, the Hi-BEHRT exceeds the state-of-the-art BEHRT 1% to 5% for area under the receiver operating characteristic (AUROC) curve and 3% to 6% for area under the precision recall (AUPRC) curve on average, and 3% to 6% (AUROC) and 3% to 11% (AUPRC) for patients with long medical history for 5-year heart failure, diabetes, chronic kidney disease, and stroke risk prediction. Additionally, because pretraining for hierarchical Transformer is not well-established, we provide an effective end-to-end contrastive pre-training strategy for Hi-BEHRT using EHR, improving its transferability on predicting clinical events with relatively small training dataset.

(2 min) Involving humans directly for the benefit of AI agents' training is getting traction thanks to several advances in reinforcement learning and human-in-the-loop learning. Humans can provide rewards to the agent, demonstrate tasks, design a curriculum, or act in the environment, but these benefits also come with architectural, functional design and engineering complexities. We present Cogment, a unifying open-source framework that introduces an actor formalism to support a variety of humans-agents collaboration typologies and training approaches. It is also scalable out of the box thanks to a distributed micro service architecture, and offers solutions to the aforementioned complexities.

(2 min) Multi-modal sentiment analysis plays an important role for providing better interactive experiences to users. Each modality in multi-modal data can provide different viewpoints or reveal unique aspects of a user's emotional state. In this work, we use text, audio and visual modalities from MOSI dataset and we propose a novel fusion technique using a multi-head attention LSTM network. Finally, we perform a classification task and evaluate its performance.

(2 min) Learning node representations on temporal graphs is a fundamental step to learn real-word dynamic graphs efficiently. Real-world graphs have the nature of continuously evolving over time, such as changing edges weights, removing and adding nodes and appearing and disappearing of edges, while previous graph representation learning methods focused generally on static graphs. We present ConvDySAT as an enhancement of DySAT, one of the state-of-the-art dynamic methods, by augmenting convolution neural networks with the self-attention mechanism, the employed method in DySAT to express the structural and temporal evolution. We conducted single-step link prediction on a communication network and rating network, Experimental results show significant performance gains for ConvDySAT over various state-of-the-art methods.

(2 min) Coronary X-ray angiography is a crucial clinical procedure for the diagnosis and treatment of coronary artery disease, which accounts for roughly 16% of global deaths every year. However, the images acquired in these procedures have low resolution and poor contrast, making lesion detection and assessment challenging. Accurate coronary artery segmentation not only helps mitigate these problems, but also allows the extraction of relevant anatomical features for further analysis by quantitative methods. Although automated segmentation of coronary arteries has been proposed before, previous approaches have used non-optimal segmentation criteria, leading to less useful results. Most methods either segment only the major vessel, discarding important information from the remaining ones, or segment the whole coronary tree based mostly on contrast information, producing a noisy output that includes vessels that are not relevant for diagnosis. We adopt a better-suited clinical criterion and segment vessels according to their clinical relevance. Additionally, we simultaneously perform catheter segmentation, which may be useful for diagnosis due to the scale factor provided by the catheter's known diameter, and is a task that has not yet been performed with good results. To derive the optimal approach, we conducted an extensive comparative study of encoder-decoder architectures trained on a combination of focal loss and a variant of generalized dice loss. Based on the EfficientNet and the UNet++ architectures, we propose a line of efficient and high-performance segmentation models using a new decoder architecture, the EfficientUNet++, whose best-performing version achieved average dice scores of 0.8904 and 0.7526 for the artery and catheter classes, respectively, and an average generalized dice score of 0.9234.

(2 min) In the text processing context, most ML models are built on word embeddings. These embeddings are themselves trained on some datasets, potentially containing sensitive data. In some cases this training is done independently, in other cases, it occurs as part of training a larger, task-specific model. In either case, it is of interest to consider membership inference attacks based on the embedding layer as a way of understanding sensitive information leakage. But, somewhat surprisingly, membership inference attacks on word embeddings and their effect in other natural language processing (NLP) tasks that use these embeddings, have remained relatively unexplored. In this work, we show that word embeddings are vulnerable to black-box membership inference attacks under realistic assumptions. Furthermore, we show that this leakage persists through two other major NLP applications: classification and text-generation, even when the embedding layer is not exposed to the attacker. We show that our MI attack achieves high attack accuracy against a classifier model and an LSTM-based language model. Indeed, our attack is a cheaper membership inference attack on text-generative models, which does not require the knowledge of the target model or any expensive training of text-generative models as shadow models.

(2 min) Bilevel optimization recently has attracted increased interest in machine learning due to its many applications such as hyper-parameter optimization and policy optimization. Although some methods recently have been proposed to solve the bilevel problems, these methods do not consider using adaptive learning rates. To fill this gap, in the paper, we propose a class of fast and effective adaptive methods for solving bilevel optimization problems that the outer problem is possibly nonconvex and the inner problem is strongly-convex. Specifically, we propose a fast single-loop BiAdam algorithm based on the basic momentum technique, which achieves a sample complexity of $\tilde{O}(\epsilon^{-4})$ for finding an $\epsilon$-stationary point. At the same time, we propose an accelerated version of BiAdam algorithm (VR-BiAdam) by using variance reduced technique, which reaches the best known sample complexity of $\tilde{O}(\epsilon^{-3})$. To further reduce computation in estimating derivatives, we propose a fast single-loop stochastic approximated BiAdam algorithm (saBiAdam) by avoiding the Hessian inverse, which still achieves a sample complexity of $\tilde{O}(\epsilon^{-4})$ without large batches. We further present an accelerated version of saBiAdam algorithm (VR-saBiAdam), which also reaches the best known sample complexity of $\tilde{O}(\epsilon^{-3})$. We apply the unified adaptive matrices to our methods as the SUPER-ADAM \citep{huang2021super}, which including many types of adaptive learning rates. Moreover, our framework can flexibly use the momentum and variance reduced techniques. In particular, we provide a useful convergence analysis framework for both the constrained and unconstrained bilevel optimization. To the best of our knowledge, we first study the adaptive bilevel optimization methods with adaptive learning rates.

(2 min) We characterize the measurement complexity of compressed sensing of signals drawn from a known prior distribution, even when the support of the prior is the entire space (rather than, say, sparse vectors). We show for Gaussian measurements and \emph{any} prior distribution on the signal, that the posterior sampling estimator achieves near-optimal recovery guarantees. Moreover, this result is robust to model mismatch, as long as the distribution estimate (e.g., from an invertible generative model) is close to the true distribution in Wasserstein distance. We implement the posterior sampling estimator for deep generative priors using Langevin dynamics, and empirically find that it produces accurate estimates with more diversity than MAP.

(2 min) This open-source book represents our attempt to make deep learning approachable, teaching readers the concepts, the context, and the code. The entire book is drafted in Jupyter notebooks, seamlessly integrating exposition figures, math, and interactive examples with self-contained code. Our goal is to offer a resource that could (i) be freely available for everyone; (ii) offer sufficient technical depth to provide a starting point on the path to actually becoming an applied machine learning scientist; (iii) include runnable code, showing readers how to solve problems in practice; (iv) allow for rapid updates, both by us and also by the community at large; (v) be complemented by a forum for interactive discussion of technical details and to answer questions.

(2 min) Unsupervised domain adaptation is used in many machine learning applications where, during training, a model has access to unlabeled data in the target domain, and a related labeled dataset. In this paper, we introduce a novel and general domain-adversarial framework. Specifically, we derive a novel generalization bound for domain adaptation that exploits a new measure of discrepancy between distributions based on a variational characterization of f-divergences. It recovers the theoretical results from Ben-David et al. (2010a) as a special case and supports divergences used in practice. Based on this bound, we derive a new algorithmic framework that introduces a key correction in the original adversarial training method of Ganin et al. (2016). We show that many regularizers and ad-hoc objectives introduced over the last years in this framework are then not required to achieve performance comparable to (if not better than) state-of-the-art domain-adversarial methods. Experimental analysis conducted on real-world natural language and computer vision datasets show that our framework outperforms existing baselines, and obtains the best results for f-divergences that were not considered previously in domain-adversarial learning.

(2 min) This paper develops an efficient procedure for designing low-complexity codebooks for precoding in a full-dimension (FD) multiple-input multiple-output (MIMO) system with a uniform planar array (UPA) antenna at the transmitter (Tx) using tensor learning. In particular, instead of using statistical channel models, we utilize a model-free data-driven approach with foundations in machine learning to generate codebooks that adapt to the surrounding propagation conditions. We use a tensor representation of the FD-MIMO channel and exploit its properties to design quantized version of the channel precoders. We find the best representation of the optimal precoder as a function of Kronecker Product (KP) of two low-dimensional precoders, respectively corresponding to the horizontal and vertical dimensions of the UPA, obtained from the tensor decomposition of the channel. We then quantize this precoder to design product codebooks such that an average loss in mutual information due to quantization of channel state information (CSI) is minimized. The key technical contribution lies in exploiting the constraints on the precoders to reduce the product codebook design problem to an unsupervised clustering problem on a Cartesian Product Grassmann manifold (CPM), where the cluster centroids form a finite-sized precoder codebook. This codebook can be found efficiently by running a $K$-means clustering on the CPM. With a suitable induced distance metric on the CPM, we show that the construction of product codebooks is equivalent to finding the optimal set of centroids on the factor manifolds corresponding to the horizontal and vertical dimensions. Simulation results are presented to demonstrate the capability of the proposed design criterion in learning the codebooks and the attractive performance of the designed codebooks.

(2 min) Large machine learning models achieve unprecedented performance on various tasks and have evolved as the go-to technique. However, deploying these compute and memory hungry models on resource constraint environments poses new challenges. In this work, we propose mathematically provable Representer Sketch, a concise set of count arrays that can approximate the inference procedure with simple hashing computations and aggregations. Representer Sketch builds upon the popular Representer Theorem from kernel literature, hence the name, providing a generic fundamental alternative to the problem of efficient inference that goes beyond the popular approach such as quantization, iterative pruning and knowledge distillation. A neural network function is transformed to its weighted kernel density representation, which can be very efficiently estimated with our sketching algorithm. Empirically, we show that Representer Sketch achieves up to 114x reduction in storage requirement and 59x reduction in computation complexity without any drop in accuracy.

(2 min) The advent of social media platforms has been a catalyst for the development of digital photography that engendered a boom in vision applications. With this motivation, we introduce a large-scale dataset termed 'Photozilla', which includes over 990k images belonging to 10 different photographic styles. The dataset is then used to train 3 classification models to automatically classify the images into the relevant style which resulted in an accuracy of ~96%. With the rapid evolution of digital photography, we have seen new types of photography styles emerging at an exponential rate. On that account, we present a novel Siamese-based network that uses the trained classification models as the base architecture to adapt and classify unseen styles with only 25 training samples. We report an accuracy of over 68% for identifying 10 other distinct types of photography styles. This dataset can be found at https://trisha025.github.io/Photozilla/

(2 min) Moving objects have special importance for Autonomous Driving tasks. Detecting moving objects can be posed as Moving Object Segmentation, by segmenting the object pixels, or Moving Object Detection, by generating a bounding box for the moving targets. In this paper, we present a Multi-Task Learning architecture, based on Transformers, to jointly perform both tasks through one network. Due to the importance of the motion features to the task, the whole setup is based on a Spatio-Temporal aggregation. We evaluate the performance of the individual tasks architecture versus the MTL setup, both with early shared encoders, and late shared encoder-decoder transformers. For the latter, we present a novel joint tasks query decoder transformer, that enables us to have tasks dedicated heads out of the shared model. To evaluate our approach, we use the KITTI MOD [29] data set. Results show1.5% mAP improvement for Moving Object Detection, and 2%IoU improvement for Moving Object Segmentation, over the individual tasks networks.

(2 min) Fact-based Visual Question Answering (FVQA), a challenging variant of VQA, requires a QA-system to include facts from a diverse knowledge graph (KG) in its reasoning process to produce an answer. Large KGs, especially common-sense KGs, are known to be incomplete, i.e., not all non-existent facts are always incorrect. Therefore, being able to reason over incomplete KGs for QA is a critical requirement in real-world applications that has not been addressed extensively in the literature. We develop a novel QA architecture that allows us to reason over incomplete KGs, something current FVQA state-of-the-art (SOTA) approaches lack due to their critical reliance on fact retrieval. We use KG Embeddings, a technique widely used for KG completion, for the downstream task of FVQA. We also employ a new image representation technique we call 'Image-as-Knowledge' to enable this capability, alongside a simple one-step CoAttention mechanism to attend to text and image during QA. Our FVQA architecture is faster during inference time, being O(m), as opposed to existing FVQA SOTA methods which are O(N log N), where m = number of vertices, N = number of edges = O(m^2). KG embeddings are shown to hold complementary information to word embeddings: a combination of both metrics permits performance comparable to SOTA methods in the standard answer retrieval task, and significantly better (26% absolute) in the proposed missing-edge reasoning task.

(2 min) Aspect-based sentiment analysis (ABSA) aims to predict the sentiment expressed in a review with respect to a given aspect. The core of ABSA is to model the interaction between the context and given aspect to extract the aspect-related information. In prior work, attention mechanisms and dependency graph networks are commonly adopted to capture the relations between the context and given aspect. And the weighted sum of context hidden states is used as the final representation fed to the classifier. However, the information related to the given aspect may be already discarded and adverse information may be retained in the context modeling processes of existing models. This problem cannot be solved by subsequent modules and there are two reasons: first, their operations are conducted on the encoder-generated context hidden states, whose value cannot change after the encoder; second, existing encoders only consider the context while not the given aspect. To address this problem, we argue the given aspect should be considered as a new clue out of context in the context modeling process. As for solutions, we design several aspect-aware context encoders based on different backbones: an aspect-aware LSTM and three aspect-aware BERTs. They are dedicated to generate aspect-aware hidden states which are tailored for ABSA task. In these aspect-aware context encoders, the semantics of the given aspect is used to regulate the information flow. Consequently, the aspect-related information can be retained and aspect-irrelevant information can be excluded in the generated hidden states. We conduct extensive experiments on several benchmark datasets with empirical analysis, demonstrating the efficacies and advantages of our proposed aspect-aware context encoders.

(2 min) Speaker diarization is a task to label audio or video recordings with classes that correspond to speaker identity, or in short, a task to identify "who spoke when". In the early years, speaker diarization algorithms were developed for speech recognition on multispeaker audio recordings to enable speaker adaptive processing. These algorithms also gained their own value as a standalone application over time to provide speaker-specific metainformation for downstream tasks such as audio retrieval. More recently, with the emergence of deep learning technology, which has driven revolutionary changes in research and practices across speech application domains, rapid advancements have been made for speaker diarization. In this paper, we review not only the historical development of speaker diarization technology but also the recent advancements in neural speaker diarization approaches. Furthermore, we discuss how speaker diarization systems have been integrated with speech recognition applications and how the recent surge of deep learning is leading the way of jointly modeling these two components to be complementary to each other. By considering such exciting technical trends, we believe that this paper is a valuable contribution to the community to provide a survey work by consolidating the recent developments with neural methods and thus facilitating further progress toward a more efficient speaker diarization.

(2 min) Learning to capture text-table alignment is essential for tasks like text-to-SQL. A model needs to correctly recognize natural language references to columns and values and to ground them in the given database schema. In this paper, we present a novel weakly supervised Structure-Grounded pretraining framework (StruG) for text-to-SQL that can effectively learn to capture text-table alignment based on a parallel text-table corpus. We identify a set of novel prediction tasks: column grounding, value grounding and column-value mapping, and leverage them to pretrain a text-table encoder. Additionally, to evaluate different methods under more realistic text-table alignment settings, we create a new evaluation set Spider-Realistic based on Spider dev set with explicit mentions of column names removed, and adopt eight existing text-to-SQL datasets for cross-database evaluation. STRUG brings significant improvement over BERT-LARGE in all settings. Compared with existing pretraining methods such as GRAPPA, STRUG achieves similar performance on Spider, and outperforms all baselines on more realistic sets. All the code and data used in this work is public available at https://aka.ms/strug.

(2 min) The flexibility of the inference process in Variational Autoencoders (VAEs) has recently led to revising traditional probabilistic topic models giving rise to Neural Topic Models (NTMs). Although these approaches have achieved significant results, surprisingly very little work has been done on how to disentangle the latent topics. Existing topic models when applied to reviews may extract topics associated with writers' subjective opinions mixed with those related to factual descriptions such as plot summaries in movie and book reviews. It is thus desirable to automatically separate opinion topics from plot/neutral ones enabling a better interpretability. In this paper, we propose a neural topic model combined with adversarial training to disentangle opinion topics from plot and neutral ones. We conduct an extensive experimental assessment introducing a new collection of movie and book reviews paired with their plots, namely MOBO dataset, showing an improved coherence and variety of topics, a consistent disentanglement rate, and sentiment classification performance superior to other supervised topic models.

(2 min) We show that Transformer encoder architectures can be massively sped up, with limited accuracy costs, by replacing the self-attention sublayers with simple linear transformations that "mix" input tokens. These linear transformations, along with standard nonlinearities in feed-forward layers, prove competent at modeling semantic relationships in several text classification tasks. Most surprisingly, we find that replacing the self-attention sublayer in a Transformer encoder with a standard, unparameterized Fourier Transform achieves 92-97% of the accuracy of BERT counterparts on the GLUE benchmark, but trains nearly seven times faster on GPUs and twice as fast on TPUs. The resulting model, FNet, also scales very efficiently to long inputs. Specifically, when compared to the "efficient" Transformers on the Long Range Arena benchmark, FNet matches the accuracy of the most accurate models, but is faster than the fastest models across all sequence lengths on GPUs (and across relatively shorter lengths on TPUs). Finally, FNet has a light memory footprint and is particularly efficient at smaller model sizes: for a fixed speed and accuracy budget, small FNet models outperform Transformer counterparts.

(2 min) The exponential rise of online social media has enabled the creation, distribution, and consumption of information at an unprecedented rate. However, it has also led to the burgeoning of various forms of online abuse. Increasing cases of online antisemitism have become one of the major concerns because of its socio-political consequences. Unlike other major forms of online abuse like racism, sexism, etc., online antisemitism has not been studied much from a machine learning perspective. To the best of our knowledge, we present the first work in the direction of automated multimodal detection of online antisemitism. The task poses multiple challenges that include extracting signals across multiple modalities, contextual references, and handling multiple aspects of antisemitism. Unfortunately, there does not exist any publicly available benchmark corpus for this critical task. Hence, we collect and label two datasets with 3,102 and 3,509 social media posts from Twitter and Gab respectively. Further, we present a multimodal deep learning system that detects the presence of antisemitic content and its specific antisemitism category using text and images from posts. We perform an extensive set of experiments on the two datasets to evaluate the efficacy of the proposed system. Finally, we also present a qualitative analysis of our study.

(2 min) Existing bias mitigation methods to reduce disparities in model outcomes across cohorts have focused on data augmentation, debiasing model embeddings, or adding fairness-based optimization objectives during training. Separately, certified word substitution robustness methods have been developed to decrease the impact of spurious features and synonym substitutions on model predictions. While their end goals are different, they both aim to encourage models to make the same prediction for certain changes in the input. In this paper, we investigate the utility of certified word substitution robustness methods to improve equality of odds and equality of opportunity on multiple text classification tasks. We observe that certified robustness methods improve fairness, and using both robustness and bias mitigation methods in training results in an improvement in both fronts

(2 min) Multi-head attention has each of the attention heads collect salient information from different parts of an input sequence, making it a powerful mechanism for sequence modeling. Multilingual and multi-domain learning are common scenarios for sequence modeling, where the key challenge is to maximize positive transfer and mitigate negative transfer across languages and domains. In this paper, we find that non-selective attention sharing is sub-optimal for achieving good generalization across all languages and domains. We further propose attention sharing strategies to facilitate parameter sharing and specialization in multilingual and multi-domain sequence modeling. Our approach automatically learns shared and specialized attention heads for different languages and domains to mitigate their interference. Evaluated in various tasks including speech recognition, text-to-text and speech-to-text translation, the proposed attention sharing strategies consistently bring gains to sequence models built upon multi-head attention. For speech-to-text translation, our approach yields an average of $+2.0$ BLEU over $13$ language directions in multilingual setting and $+2.0$ BLEU over $3$ domains in multi-domain setting.

(2 min) Natural reading orders of words are crucial for information extraction from form-like documents. Despite recent advances in Graph Convolutional Networks (GCNs) on modeling spatial layout patterns of documents, they have limited ability to capture reading orders of given word-level node representations in a graph. We propose Reading Order Equivariant Positional Encoding (ROPE), a new positional encoding technique designed to apprehend the sequential presentation of words in documents. ROPE generates unique reading order codes for neighboring words relative to the target word given a word-level graph connectivity. We study two fundamental document entity extraction tasks including word labeling and word grouping on the public FUNSD dataset and a large-scale payment dataset. We show that ROPE consistently improves existing GCNs with a margin up to 8.4% F1-score.

(2 min) In this paper, we explore the ability to model and infer personality types of opponents, predict their responses, and use this information to adapt a dialog agent's high-level strategy in negotiation tasks. Inspired by the idea of incorporating a theory of mind (ToM) into machines, we introduce a probabilistic formulation to encapsulate the opponent's personality type during both learning and inference. We test our approach on the CraigslistBargain dataset and show that our method using ToM inference achieves a 20% higher dialog agreement rate compared to baselines on a mixed population of opponents. We also find that our model displays diverse negotiation behavior with different types of opponents.

(2 min) Automatic generation of high-quality commit messages for code commits can substantially facilitate software developers' works and coordination. However, the semantic gap between source code and natural language poses a major challenge for the task. Several studies have been proposed to alleviate the challenge but none explicitly involves code contextual information during commit message generation. Specifically, existing research adopts static embedding for code tokens, which maps a token to the same vector regardless of its context. In this paper, we propose a novel Contextualized code representation learning strategy for commit message Generation (CoreGen). CoreGen first learns contextualized code representations which exploit the contextual information behind code commit sequences. The learned representations of code commits built upon Transformer are then fine-tuned for downstream commit message generation. Experiments on the benchmark dataset demonstrate the superior effectiveness of our model over the baseline models with at least 28.18% improvement in terms of BLEU-4 score. Furthermore, we also highlight the future opportunities in training contextualized code representations on larger code corpus as a solution to low-resource tasks and adapting the contextualized code representation framework to other code-to-text generation tasks.

(3 min) Nowadays, there are many applications of text mining over corpora from different languages. However, most of them are based on texts in prose, lacking applications that work with poetry texts. An example of an application of text mining in poetry is the usage of features derived from their individual words in order to capture the lexical, sublexical and interlexical meaning, and infer the General Affective Meaning (GAM) of the text. However, even though this proposal has been proved as useful for poetry in some languages, there is a lack of studies for both Spanish poetry and for highly-structured poetic compositions such as sonnets. This article presents a study over an annotated corpus of Spanish sonnets, in order to analyse if it is possible to build features from their individual words for predicting their GAM. The purpose of this is to model sonnets at an affective level. The article also analyses the relationship between the GAM of the sonnets and the content itself. For this, we consider the content from a psychological perspective, identifying with tags when a sonnet is related to a specific term. Then, we study how GAM changes according to each of those psychological terms. The corpus used contains 274 Spanish sonnets from authors of different centuries, from 15th to 19th. This corpus was annotated by different domain experts. The experts annotated the poems with affective and lexico-semantic features, as well as with domain concepts that belong to psychology. Thanks to this, the corpus of sonnets can be used in different applications, such as poetry recommender systems, personality text mining studies of the authors, or the usage of poetry for therapeutic purposes.

(2 min) By computing the rank correlation between attention weights and feature-additive explanation methods, previous analyses either invalidate or support the role of attention-based explanations as a faithful and plausible measure of salience. To investigate whether this approach is appropriate, we compare LIME, Integrated Gradients, DeepLIFT, Grad-SHAP, Deep-SHAP, and attention-based explanations, applied to two neural architectures trained on single- and pair-sequence language tasks. In most cases, we find that none of our chosen methods agree. Based on our empirical observations and theoretical objections, we conclude that rank correlation does not measure the quality of feature-additive methods. Practitioners should instead use the numerous and rigorous diagnostic methods proposed by the community.

(2 min) Fact Verification requires fine-grained natural language inference capability that finds subtle clues to identify the syntactical and semantically correct but not well-supported claims. This paper presents Kernel Graph Attention Network (KGAT), which conducts more fine-grained fact verification with kernel-based attentions. Given a claim and a set of potential evidence sentences that form an evidence graph, KGAT introduces node kernels, which better measure the importance of the evidence node, and edge kernels, which conduct fine-grained evidence propagation in the graph, into Graph Attention Networks for more accurate fact verification. KGAT achieves a 70.38% FEVER score and significantly outperforms existing fact verification models on FEVER, a large-scale benchmark for fact verification. Our analyses illustrate that, compared to dot-product attentions, the kernel-based attention concentrates more on relevant evidence sentences and meaningful clues in the evidence graph, which is the main source of KGAT's effectiveness.

(2 min) Self-supervised learning (SSL) has proven vital for advancing research in natural language processing (NLP) and computer vision (CV). The paradigm pretrains a shared model on large volumes of unlabeled data and achieves state-of-the-art (SOTA) for various tasks with minimal adaptation. However, the speech processing community lacks a similar setup to systematically explore the paradigm. To bridge this gap, we introduce Speech processing Universal PERformance Benchmark (SUPERB). SUPERB is a leaderboard to benchmark the performance of a shared model across a wide range of speech processing tasks with minimal architecture changes and labeled data. Among multiple usages of the shared model, we especially focus on extracting the representation learned from SSL due to its preferable re-usability. We present a simple framework to solve SUPERB tasks by learning task-specialized lightweight prediction heads on top of the frozen shared model. Our results demonstrate that the framework is promising as SSL representations show competitive generalizability and accessibility across SUPERB tasks. We release SUPERB as a challenge with a leaderboard and a benchmark toolkit to fuel the research in representation learning and general speech processing.

(2 min) Named entity recognition (NER) is a widely studied task in natural language processing. Recently, a growing number of studies have focused on the nested NER. The span-based methods, considering the entity recognition as a span classification task, can deal with nested entities naturally. But they suffer from the huge search space and the lack of interactions between entities. To address these issues, we propose a novel sequence-to-set neural network for nested NER. Instead of specifying candidate spans in advance, we provide a fixed set of learnable vectors to learn the patterns of the valuable spans. We utilize a non-autoregressive decoder to predict the final set of entities in one pass, in which we are able to capture dependencies between entities. Compared with the sequence-to-sequence method, our model is more suitable for such unordered recognition task as it is insensitive to the label order. In addition, we utilize the loss function based on bipartite matching to compute the overall training loss. Experimental results show that our proposed model achieves state-of-the-art on three nested NER corpora: ACE 2004, ACE 2005 and KBP 2017. The code is available at https://github.com/zqtan1024/sequence-to-set.

(2 min) With recent advances in distantly supervised (DS) relation extraction (RE), considerable attention is attracted to leverage multi-instance learning (MIL) to distill high-quality supervision from the noisy DS. Here, we go beyond label noise and identify the key bottleneck of DS-MIL to be its low data utilization: as high-quality supervision being refined by MIL, MIL abandons a large amount of training instances, which leads to a low data utilization and hinders model training from having abundant supervision. In this paper, we propose collaborative adversarial training to improve the data utilization, which coordinates virtual adversarial training (VAT) and adversarial training (AT) at different levels. Specifically, since VAT is label-free, we employ the instance-level VAT to recycle instances abandoned by MIL. Besides, we deploy AT at the bag-level to unleash the full potential of the high-quality supervision got by MIL. Our proposed method brings consistent improvements (~ 5 absolute AUC score) to the previous state of the art, which verifies the importance of the data utilization issue and the effectiveness of our method.

(2 min) We present an approach for unsupervised learning of speech representation disentangling contents and styles. Our model consists of: (1) a local encoder that captures per-frame information; (2) a global encoder that captures per-utterance information; and (3) a conditional decoder that reconstructs speech given local and global latent variables. Our experiments show that (1) the local latent variables encode speech contents, as reconstructed speech can be recognized by ASR with low word error rates (WER), even with a different global encoding; (2) the global latent variables encode speaker style, as reconstructed speech shares speaker identity with the source utterance of the global encoding. Additionally, we demonstrate an useful application from our pre-trained model, where we can train a speaker recognition model from the global latent variables and achieve high accuracy by fine-tuning with as few data as one label per speaker.

(2 min) Despite pre-trained language models such as BERT have achieved appealing performance in a wide range of natural language processing tasks, they are computationally expensive to be deployed in real-time applications. A typical method is to adopt knowledge distillation to compress these large pre-trained models (teacher models) to small student models. However, for a target domain with scarce training data, the teacher can hardly pass useful knowledge to the student, which yields performance degradation for the student models. To tackle this problem, we propose a method to learn to augment for data-scarce domain BERT knowledge distillation, by learning a cross-domain manipulation scheme that automatically augments the target with the help of resource-rich source domains. Specifically, the proposed method generates samples acquired from a stationary distribution near the target data and adopts a reinforced selector to automatically refine the augmentation strategy according to the performance of the student. Extensive experiments demonstrate that the proposed method significantly outperforms state-of-the-art baselines on four different tasks, and for the data-scarce domains, the compressed student models even perform better than the original large teacher model, with much fewer parameters (only ${\sim}13.3\%$) when only a few labeled examples available.

(2 min) Lawyers and judges spend a large amount of time researching the proper legal authority to cite while drafting decisions. In this paper, we develop a citation recommendation tool that can help improve efficiency in the process of opinion drafting. We train four types of machine learning models, including a citation-list based method (collaborative filtering) and three context-based methods (text similarity, BiLSTM and RoBERTa classifiers). Our experiments show that leveraging local textual context improves recommendation, and that deep neural models achieve decent performance. We show that non-deep text-based methods benefit from access to structured case metadata, but deep models only benefit from such access when predicting from context of insufficient length. We also find that, even after extensive training, RoBERTa does not outperform a recurrent neural model, despite its benefits of pretraining. Our behavior analysis of the RoBERTa model further shows that predictive performance is stable across time and citation classes.

(2 min) Recent advancements in deep learning have created many opportunities to solve real-world problems that remained unsolved for more than a decade. Automatic caption generation is a major research field, and the research community has done a lot of work on it in most common languages like English. Urdu is the national language of Pakistan and also much spoken and understood in the sub-continent region of Pakistan-India, and yet no work has been done for Urdu language caption generation. Our research aims to fill this gap by developing an attention-based deep learning model using techniques of sequence modeling specialized for the Urdu language. We have prepared a dataset in the Urdu language by translating a subset of the "Flickr8k" dataset containing 700 'man' images. We evaluate our proposed technique on this dataset and show that it can achieve a BLEU score of 0.83 in the Urdu language. We improve on the previous state-of-the-art by using better CNN architectures and optimization techniques. Furthermore, we provide a discussion on how the generated captions can be made correct grammar-wise.

(2 min) Cross-lingual representations have the potential to make NLP techniques available to the vast majority of languages in the world. However, they currently require large pretraining corpora or access to typologically similar languages. In this work, we address these obstacles by removing language identity signals from multilingual embeddings. We examine three approaches for this: (i) re-aligning the vector spaces of target languages (all together) to a pivot source language; (ii) removing language-specific means and variances, which yields better discriminativeness of embeddings as a by-product; and (iii) increasing input similarity across languages by removing morphological contractions and sentence reordering. We evaluate on XNLI and reference-free MT across 19 typologically diverse languages. Our findings expose the limitations of these approaches -- unlike vector normalization, vector space re-alignment and text normalization do not achieve consistent gains across encoders and languages. Due to the approaches' additive effects, their combination decreases the cross-lingual transfer gap by 8.9 points (m-BERT) and 18.2 points (XLM-R) on average across all tasks and languages, however. Our code and models are publicly available.

(2 min) In this paper we present an efficient method for training models for speaker recognition using small or under-resourced datasets. This method requires less data than other SOTA (State-Of-The-Art) methods, e.g. the Angular Prototypical and GE2E loss functions, while achieving similar results to those methods. This is done using the knowledge of the reconstruction of a phoneme in the speaker's voice. For this purpose, a new dataset was built, composed of 40 male speakers, who read sentences in Portuguese, totaling approximately 3h. We compare the three best architectures trained using our method to select the best one, which is the one with a shallow architecture. Then, we compared this model with the SOTA method for the speaker recognition task: the Fast ResNet-34 trained with approximately 2,000 hours, using the loss functions Angular Prototypical and GE2E. Three experiments were carried out with datasets in different languages. Among these three experiments, our model achieved the second best result in two experiments and the best result in one of them. This highlights the importance of our method, which proved to be a great competitor to SOTA speaker recognition models, with 500x less data and a simpler approach.

(2 min) Dialogue relation extraction (DRE) aims to detect the relation between two entities mentioned in a multi-party dialogue. It plays an important role in constructing knowledge graphs from conversational data increasingly abundant on the internet and facilitating intelligent dialogue system development. The prior methods of DRE do not meaningfully leverage speaker information-they just prepend the utterances with the respective speaker names. Thus, they fail to model the crucial inter-speaker relations that may give additional context to relevant argument entities through pronouns and triggers. We, however, present a graph attention network-based method for DRE where a graph, that contains meaningfully connected speaker, entity, entity-type, and utterance nodes, is constructed. This graph is fed to a graph attention network for context propagation among relevant nodes, which effectively captures the dialogue context. We empirically show that this graph-based approach quite effectively captures the relations between different entity pairs in a dialogue as it outperforms the state-of-the-art approaches by a significant margin on the benchmark dataset DialogRE. Our code is released at: https://github.com/declare-lab/dialog-HGAT

(3 min) The Grammar of Institutions, or Institutional Grammar, is an established approach to encode policy information in terms of institutional statements based on a set of pre-defined syntactic components. This codebook provides coding guidelines for a revised version of the Institutional Grammar, the Institutional Grammar 2.0 (IG 2.0). IG 2.0 is a specification that aims at facilitating the encoding of policy to meet varying analytical objectives. To this end, it revises the grammar with respect to comprehensiveness, flexibility, and specificity by offering multiple levels of expressiveness (IG Core, IG Extended, IG Logico). In addition to the encoding of regulative statements, it further introduces the encoding of constitutive institutional statements, as well as statements that exhibit both constitutive and regulative characteristics. Introducing those aspects, the codebook initially covers fundamental concepts of IG 2.0, before providing an overview of pre-coding steps relevant for document preparation. Detailed coding guidelines are provided for both regulative and constitutive statements across all levels of expressiveness, along with the encoding guidelines for statements of mixed form -- hybrid and polymorphic institutional statements. The document further provides an overview of taxonomies used in the encoding process and referred to throughout the codebook. The codebook concludes with a summary and discussion of relevant considerations to facilitate the coding process. An initial Reader's Guide helps the reader tailor the content to her interest. Note that this codebook specifically focuses on operational aspects of IG 2.0 in the context of policy coding. Links to additional resources such as the underlying scientific literature (that offers a comprehensive treatment of the underlying theoretical concepts) are referred to in the concluding section of the codebook.

(2 min) In recent years, the size of pre-trained language models (PLMs) has grown by leaps and bounds. However, efficiency issues of these large-scale PLMs limit their utilization in real-world scenarios. We present a suite of cost-effective techniques for the use of PLMs to deal with the efficiency issues of pre-training, fine-tuning, and inference. (1) We introduce knowledge inheritance to accelerate the pre-training process by exploiting existing PLMs instead of training models from scratch. (2) We explore the best practice of prompt tuning with large-scale PLMs. Compared with conventional fine-tuning, prompt tuning significantly reduces the number of task-specific parameters. (3) We implement a new inference toolkit, namely InfMoE, for using large-scale PLMs with limited computational resources. Based on our cost-effective pipeline, we pre-train two models: an encoder-decoder bilingual model with 11 billion parameters (CPM-2) and its corresponding MoE version with 198 billion parameters. In our experiments, we compare CPM-2 with mT5 on downstream tasks. Experimental results show that CPM-2 has excellent general language intelligence. Moreover, we validate the efficiency of InfMoE when conducting inference of large-scale models having tens of billions of parameters on a single GPU. All source code and model parameters are available at https://github.com/TsinghuaAI/CPM.

(2 min) Assessing an AI agent that can converse in human language and understand visual content is challenging. Generation metrics, such as BLEU scores favor correct syntax over semantics. Hence a discriminative approach is often used, where an agent ranks a set of candidate options. The mean reciprocal rank (MRR) metric evaluates the model performance by taking into account the rank of a single human-derived answer. This approach, however, raises a new challenge: the ambiguity and synonymy of answers, for instance, semantic equivalence (e.g., `yeah' and `yes'). To address this, the normalized discounted cumulative gain (NDCG) metric has been used to capture the relevance of all the correct answers via dense annotations. However, the NDCG metric favors the usually applicable uncertain answers such as `I don't know. Crafting a model that excels on both MRR and NDCG metrics is challenging. Ideally, an AI agent should answer a human-like reply and validate the correctness of any answer. To address this issue, we describe a two-step non-parametric ranking approach that can merge strong MRR and NDCG models. Using our approach, we manage to keep most MRR state-of-the-art performance (70.41% vs. 71.24%) and the NDCG state-of-the-art performance (72.16% vs. 75.35%). Moreover, our approach won the recent Visual Dialog 2020 challenge. Source code is available at https://github.com/idansc/mrr-ndcg.

(2 min) Predicting the next utterance in dialogue is contingent on encoding of users' input text to generate appropriate and relevant response in data-driven approaches. Although the semantic and syntactic quality of the language generated is evaluated, more often than not, the encoded representation of input is not evaluated. As the representation of the encoder is essential for predicting the appropriate response, evaluation of encoder representation is a challenging yet important problem. In this work, we showcase evaluating the text generated through human or automatic metrics is not sufficient to appropriately evaluate soundness of the language understanding of dialogue models and, to that end, propose a set of probe tasks to evaluate encoder representation of different language encoders commonly used in dialogue models. From experiments, we observe that some of the probe tasks are easier and some are harder for even sophisticated model architectures to learn. And, through experiments we observe that RNN based architectures have lower performance on automatic metrics on text generation than transformer model but perform better than the transformer model on the probe tasks indicating that RNNs might preserve task information better than the Transformers.

(2 min) Question generation (QG) is to generate natural and grammatical questions that can be answered by a specific answer for a given context. Previous sequence-to-sequence models suffer from a problem that asking high-quality questions requires commonsense knowledge as backgrounds, which in most cases can not be learned directly from training data, resulting in unsatisfactory questions deprived of knowledge. In this paper, we propose a multi-task learning framework to introduce commonsense knowledge into question generation process. We first retrieve relevant commonsense knowledge triples from mature databases and select triples with the conversion information from source context to question. Based on these informative knowledge triples, we design two auxiliary tasks to incorporate commonsense knowledge into the main QG model, where one task is Concept Relation Classification and the other is Tail Concept Generation. Experimental results on SQuAD show that our proposed methods are able to noticeably improve the QG performance on both automatic and human evaluation metrics, demonstrating that incorporating external commonsense knowledge with multi-task learning can help the model generate human-like and high-quality questions.

(2 min) Calligraphy is an essential part of the Arabic heritage and culture. It has been used in the past for the decoration of houses and mosques. Usually, such calligraphy is designed manually by experts with aesthetic insights. In the past few years, there has been a considerable effort to digitize such type of art by either taking a photo of decorated buildings or drawing them using digital devices. The latter is considered an online form where the drawing is tracked by recording the apparatus movement, an electronic pen for instance, on a screen. In the literature, there are many offline datasets collected with a diversity of Arabic styles for calligraphy. However, there is no available online dataset for Arabic calligraphy. In this paper, we illustrate our approach for the collection and annotation of an online dataset for Arabic calligraphy called Calliar that consists of 2,500 sentences. Calliar is annotated for stroke, character, word and sentence level prediction.

(2 min) Although emotions are universal concepts, transferring the different shades of emotion from one language to another may not always be straightforward for human translators, let alone for machine translation systems. Moreover, the cognitive states are established by verbal explanations of experience which is shaped by both the verbal and cultural contexts. There are a number of verbal contexts where expression of emotions constitutes the pivotal component of the message. This is particularly true for User-Generated Content (UGC) which can be in the form of a review of a product or a service, a tweet, or a social media post. Recently, it has become common practice for multilingual websites such as Twitter to provide an automatic translation of UGC to reach out to their linguistically diverse users. In such scenarios, the process of translating the user's emotion is entirely automatic with no human intervention, neither for post-editing nor for accuracy checking. In this research, we assess whether automatic translation tools can be a successful real-life utility in transferring emotion in user-generated multilingual data such as tweets. We show that there are linguistic phenomena specific of Twitter data that pose a challenge in translation of emotions in different languages. We summarise these challenges in a list of linguistic features and show how frequent these features are in different language pairs. We also assess the capacity of commonly used methods for evaluating the performance of an MT system with respect to the preservation of emotion in the source text.

(2 min) In this paper, we explore various multilingual and Russian pre-trained transformer-based models for the Dialogue Evaluation 2021 shared task on headline selection. Our experiments show that the combined approach is superior to individual multilingual and monolingual models. We present an analysis of a number of ways to obtain sentence embeddings and learn a ranking model on top of them. We achieve the result of 87.28% and 86.60% accuracy for the public and private test sets respectively.

(2 min) Text-style transfer aims to convert text given in one domain into another by paraphrasing the sentence or substituting the keywords without altering the content. By necessity, state-of-the-art methods have evolved to accommodate nonparallel training data, as it is frequently the case there are multiple data sources of unequal size, with a mixture of labeled and unlabeled sentences. Moreover, the inherent style defined within each source might be distinct. A generic bidirectional (e.g., formal $\Leftrightarrow$ informal) style transfer regardless of different groups may not generalize well to different applications. In this work, we developed a task adaptive meta-learning framework that can simultaneously perform a multi-pair text-style transfer using a single model. The proposed method can adaptively balance the difference of meta-knowledge across multiple tasks. Results show that our method leads to better quantitative performance as well as coherent style variations. Common challenges of unbalanced data and mismatched domains are handled well by this method.

(2 min) Sentiment and lexical analyses are widely used to detect depression or anxiety disorders. It has been documented that there are significant differences in the language used by a person with emotional disorders in comparison to a healthy individual. Still, the effectiveness of these lexical approaches could be improved further because the current analysis focuses on what the social media entries are about, and not how they are written. In this study, we focus on aspects in which these short texts are similar to each other, and how they were created. We present an innovative approach to the depression screening problem by applying Collgram analysis, which is a known effective method of obtaining linguistic information from texts. We compare these results with sentiment analysis based on the BERT architecture. Finally, we create a hybrid model achieving a diagnostic accuracy of 71%.

(2 min) Language models can generate harmful and biased outputs and exhibit undesirable behavior. We propose a Process for Adapting Language Models to Society (PALMS) with Values-Targeted Datasets, an iterative process to significantly change model behavior by crafting and fine-tuning on a dataset that reflects a predetermined set of target values. We evaluate our process using three metrics: quantitative metrics with human evaluations that score output adherence to a target value, and toxicity scoring on outputs; and qualitative metrics analyzing the most common word associated with a given social category. Through each iteration, we add additional training dataset examples based on observed shortcomings from evaluations. PALMS performs significantly better on all metrics compared to baseline and control models for a broad range of GPT-3 language model sizes without compromising capability integrity. We find that the effectiveness of PALMS increases with model size. We show that significantly adjusting language model behavior is feasible with a small, hand-curated dataset.

(2 min) Select-then-compress is a popular hybrid, framework for text summarization due to its high efficiency. This framework first selects salient sentences and then independently condenses each of the selected sentences into a concise version. However, compressing sentences separately ignores the context information of the document, and is therefore prone to delete salient information. To address this limitation, we propose a novel condense-then-select framework for text summarization. Our framework first concurrently condenses each document sentence. Original document sentences and their compressed versions then become the candidates for extraction. Finally, an extractor utilizes the context information of the document to select candidates and assembles them into a summary. If salient information is deleted during condensing, the extractor can select an original sentence to retain the information. Thus, our framework helps to avoid the loss of salient information, while preserving the high efficiency of sentence-level compression. Experiment results on the CNN/DailyMail, DUC-2002, and Pubmed datasets demonstrate that our framework outperforms the select-then-compress framework and other strong baselines.

(2 min) Existing pre-trained models for knowledge-graph-to-text (KG-to-text) generation simply fine-tune text-to-text pre-trained models such as BART or T5 on KG-to-text datasets, which largely ignore the graph structure during encoding and lack elaborate pre-training tasks to explicitly model graph-text alignments. To tackle these problems, we propose a graph-text joint representation learning model called JointGT. During encoding, we devise a structure-aware semantic aggregation module which is plugged into each Transformer layer to preserve the graph structure. Furthermore, we propose three new pre-training tasks to explicitly enhance the graph-text alignment including respective text / graph reconstruction, and graph-text alignment in the embedding space via Optimal Transport. Experiments show that JointGT obtains new state-of-the-art performance on various KG-to-text datasets.

(2 min) Neural models trained for next utterance generation in dialogue task learn to mimic the n-gram sequences in the training set with training objectives like negative log-likelihood (NLL) or cross-entropy. Such commonly used training objectives do not foster generating alternate responses to a context. But, the effects of minimizing an alternate training objective that fosters a model to generate alternate response and score it on semantic similarity has not been well studied. We hypothesize that a language generation model can improve on its diversity by learning to generate alternate text during training and minimizing a semantic loss as an auxiliary objective. We explore this idea on two different sized data sets on the task of next utterance generation in goal oriented dialogues. We make two observations (1) minimizing a semantic objective improved diversity in responses in the smaller data set (Frames) but only as-good-as minimizing the NLL in the larger data set (MultiWoZ) (2) large language model embeddings can be more useful as a semantic loss objective than as initialization for token embeddings.

(2 min) We present TweeNLP, a one-stop portal that organizes Twitter's natural language processing (NLP) data and builds a visualization and exploration platform. It curates 19,395 tweets (as of April 2021) from various NLP conferences and general NLP discussions. It supports multiple features such as TweetExplorer to explore tweets by topics, visualize insights from Twitter activity throughout the organization cycle of conferences, discover popular research papers and researchers. It also builds a timeline of conference and workshop submission deadlines. We envision TweeNLP to function as a collective memory unit for the NLP community by integrating the tweets pertaining to research papers with the NLPExplorer scientific literature search engine. The current system is hosted at this http URL .

(2 min) Compositional generalization is the ability to generalize systematically to a new data distribution by combining known components. Although humans seem to have a great ability to generalize compositionally, state-of-the-art neural models struggle to do so. In this work, we study compositional generalization in classification tasks and present two main contributions. First, we study ways to convert a natural language sequence-to-sequence dataset to a classification dataset that also requires compositional generalization. Second, we show that providing structural hints (specifically, providing parse trees and entity links as attention masks for a Transformer model) helps compositional generalization.

(2 min) Why should we trust the detections of deep neural networks for manipulated faces? Understanding the reasons is important for users in improving the fairness, reliability, privacy and trust of the detection models. In this work, we propose an interpretable face manipulation detection approach to achieve the trustworthy and accurate inference. The approach could make the face manipulation detection process transparent by embedding the feature whitening module. This module aims to whiten the internal working mechanism of deep networks through feature decorrelation and feature constraint. The experimental results demonstrate that our proposed approach can strike a balance between the detection accuracy and the model interpretability.

(2 min) Unsupervised domain adaptive classification intends to improve theclassification performance on unlabeled target domain. To alleviate the adverse effect of domain shift, many approaches align the source and target domains in the feature space. However, a feature is usually taken as a whole for alignment without explicitly making domain alignment proactively serve the classification task, leading to sub-optimal solution. What sub-feature should be aligned for better adaptation is under-explored. In this paper, we propose an effective Task-oriented Alignment (ToAlign) for unsupervised domain adaptation (UDA). We study what features should be aligned across domains and propose to make the domain alignment proactively serve classification by performing feature decomposition and alignment under the guidance of the prior knowledge induced from the classification taskitself. Particularly, we explicitly decompose a feature in the source domain intoa task-related/discriminative feature that should be aligned, and a task-irrelevant feature that should be avoided/ignored, based on the classification meta-knowledge. Extensive experimental results on various benchmarks (e.g., Office-Home, Visda-2017, and DomainNet) under different domain adaptation settings demonstrate theeffectiveness of ToAlign which helps achieve the state-of-the-art performance.

(2 min) Previous studies dominantly target at self-supervised learning on real-valued networks and have achieved many promising results. However, on the more challenging binary neural networks (BNNs), this task has not yet been fully explored in the community. In this paper, we focus on this more difficult scenario: learning networks where both weights and activations are binary, meanwhile, without any human annotated labels. We observe that the commonly used contrastive objective is not satisfying on BNNs for competitive accuracy, since the backbone network contains relatively limited capacity and representation ability. Hence instead of directly applying existing self-supervised methods, which cause a severe decline in performance, we present a novel guided learning paradigm from real-valued to distill binary networks on the final prediction distribution, to minimize the loss and obtain desirable accuracy. Our proposed method can boost the simple contrastive learning baseline by an absolute gain of 5.5~15% on BNNs. We further reveal that it is difficult for BNNs to recover the similar predictive distributions as real-valued models when training without labels. Thus, how to calibrate them is key to address the degradation in performance. Extensive experiments are conducted on the large-scale ImageNet and downstream datasets. Our method achieves substantial improvement over the simple contrastive learning baseline, and is even comparable to many mainstream supervised BNN methods. Code is available at https://github.com/szq0214/S2-BNN.

(2 min) Various fonts give different impressions, such as legible, rough, and comic-text.This paper aims to analyze the correlation between the local shapes, or parts, and the impression of fonts. By focusing on local shapes instead of the whole letter shape, we can realize letter-shape independent and more general analysis. The analysis is performed by newly combining SIFT and DeepSets, to extract an arbitrary number of essential parts from a particular font and aggregate them to infer the font impressions by nonlinear regression. Our qualitative and quantitative analyses prove that (1)fonts with similar parts have similar impressions, (2)many impressions, such as legible and rough, largely depend on specific parts, (3)several impressions are very irrelevant to parts.

(2 min) As a 3D counterpart of object classification in images, object point cloud classification is fundamental to 3D scene understanding, and has drawn great research attention since the release of benchmarking datasets, such as the ModelNet and the ShapeNet. These benchmarks assume point clouds covering complete surfaces of object instances, for which plenty of high-performing methods have been developed. However, their settings deviate from those often met in practice, where, due to (self-)occlusion, a point cloud covering partial surface of an object is captured from an arbitrary view. We show in this paper that performance of existing point cloud classification methods drops drastically under the considered practical single-view, partial setting; the phenomenon is consistent with the observation that semantic category of a partial object surface is less ambiguous only when its distribution on the whole surface is clearly specified. To this end, we argue for a single-view, partial setting where supervised learning of object pose estimation should be accompanied with classification. Technically, we propose a baseline method of Pose-Accompanied Point cloud classification Network (PAPNet); built upon SE(3)-equivariant convolutions, the PAPNet learns intermediate pose transformations for equivariant features defined on vector fields, which makes the subsequent classification easier (ideally) in the category-level, canonical pose. We adapt existing ModelNet40 and ScanNet datasets on point set classification to the introduced single-view, partial setting to verify our hypothesis. Thorough experiments confirm the necessity of object pose estimation; our PAPNet also outperforms existing methods greatly on the new benchmarks.

(2 min) Deep neural networks that yield human interpretable decisions by architectural design have lately become an increasingly popular alternative to post hoc interpretation of traditional black-box models. Among these networks, the arguably most widespread approach is so-called prototype learning, where similarities to learned latent prototypes serve as the basis of classifying an unseen data point. In this work, we point to an important shortcoming of such approaches. Namely, there is a semantic gap between similarity in latent space and similarity in input space, which can corrupt interpretability. We design two experiments that exemplify this issue on the so-called ProtoPNet. Specifically, we find that this network's interpretability mechanism can be led astray by intentionally crafted or even JPEG compression artefacts, which can produce incomprehensible decisions. We argue that practitioners ought to have this shortcoming in mind when deploying prototype-based models in practice.

(2 min) We present a method that learns a spatiotemporal neural irradiance field for dynamic scenes from a single video. Our learned representation enables free-viewpoint rendering of the input video. Our method builds upon recent advances in implicit representations. Learning a spatiotemporal irradiance field from a single video poses significant challenges because the video contains only one observation of the scene at any point in time. The 3D geometry of a scene can be legitimately represented in numerous ways since varying geometry (motion) can be explained with varying appearance and vice versa. We address this ambiguity by constraining the time-varying geometry of our dynamic scene representation using the scene depth estimated from video depth estimation methods, aggregating contents from individual frames into a single global representation. We provide an extensive quantitative evaluation and demonstrate compelling free-viewpoint rendering results.

(2 min) Most existing approaches for visual localization either need a detailed 3D model of the environment or, in the case of learning-based methods, must be retrained for each new scene. This can either be very expensive or simply impossible for large, unknown environments, for example in search-and-rescue scenarios. Although there are learning-based approaches that operate scene-agnostically, the generalization capability of these methods is still outperformed by classical approaches. In this paper, we present an approach that can generalize to new scenes by applying specific changes to the model architecture, including an extended regression part, the use of hierarchical correlation layers, and the exploitation of scale and uncertainty information. Our approach outperforms the 5-point algorithm using SIFT features on equally big images and additionally surpasses all previous learning-based approaches that were trained on different data. It is also superior to most of the approaches that were specifically trained on the respective scenes. We also evaluate our approach in a scenario where only very few reference images are available, showing that under such more realistic conditions our learning-based approach considerably exceeds both existing learning-based and classical methods.

(2 min) We present a novel approach to the detection and characterization of edges, ridges, and blobs in two-dimensional images which exploits the symmetry properties of directionally sensitive analyzing functions in multiscale systems that are constructed in the framework of alpha-molecules. The proposed feature detectors are inspired by the notion of phase congruency, stable in the presence of noise, and by definition invariant to changes in contrast. We also show how the behavior of coefficients corresponding to differently scaled and oriented analyzing functions can be used to obtain a comprehensive characterization of the geometry of features in terms of local tangent directions, widths, and heights. The accuracy and robustness of the proposed measures are validated and compared to various state-of-the-art algorithms in extensive numerical experiments in which we consider sets of clean and distorted synthetic images that are associated with reliable ground truths. To further demonstrate the applicability, we show how the proposed ridge measure can be used to detect and characterize blood vessels in digital retinal images and how the proposed blob measure can be applied to automatically count the number of cell colonies in a Petri dish.

(2 min) Security-sensitive applications that rely on Deep Neural Networks (DNNs) are vulnerable to small perturbations that are crafted to generate Adversarial Examples(AEs). The AEs are imperceptible to humans and cause DNN to misclassify them. Many defense and detection techniques have been proposed. Model's confidences and Dropout, as a popular way to estimate the model's uncertainty, have been used for AE detection but they showed limited success against black- and gray-box attacks. Moreover, the state-of-the-art detection techniques have been designed for specific attacks or broken by others, need knowledge about the attacks, are not consistent, increase model parameters overhead, are time-consuming, or have latency in inference time. To trade off these factors, we revisit the model's uncertainty and confidences and propose a novel unsupervised ensemble AE detection mechanism that 1) uses the uncertainty method called SelectiveNet, 2) processes model layers outputs, i.e.feature maps, to generate new confidence probabilities. The detection method is called Selective and Feature based Adversarial Detection (SFAD). Experimental results show that the proposed approach achieves better performance against black- and gray-box attacks than the state-of-the-art methods and achieves comparable performance against white-box attacks. Moreover, results show that SFAD is fully robust against High Confidence Attacks (HCAs) for MNIST and partially robust for CIFAR10 datasets.

(2 min) Assessing an AI agent that can converse in human language and understand visual content is challenging. Generation metrics, such as BLEU scores favor correct syntax over semantics. Hence a discriminative approach is often used, where an agent ranks a set of candidate options. The mean reciprocal rank (MRR) metric evaluates the model performance by taking into account the rank of a single human-derived answer. This approach, however, raises a new challenge: the ambiguity and synonymy of answers, for instance, semantic equivalence (e.g., `yeah' and `yes'). To address this, the normalized discounted cumulative gain (NDCG) metric has been used to capture the relevance of all the correct answers via dense annotations. However, the NDCG metric favors the usually applicable uncertain answers such as `I don't know. Crafting a model that excels on both MRR and NDCG metrics is challenging. Ideally, an AI agent should answer a human-like reply and validate the correctness of any answer. To address this issue, we describe a two-step non-parametric ranking approach that can merge strong MRR and NDCG models. Using our approach, we manage to keep most MRR state-of-the-art performance (70.41% vs. 71.24%) and the NDCG state-of-the-art performance (72.16% vs. 75.35%). Moreover, our approach won the recent Visual Dialog 2020 challenge. Source code is available at https://github.com/idansc/mrr-ndcg.

(2 min) Image captioning has recently demonstrated impressive progress largely owing to the introduction of neural network algorithms trained on curated dataset like MS-COCO. Often work in this field is motivated by the promise of deployment of captioning systems in practical applications. However, the scarcity of data and contexts in many competition datasets renders the utility of systems trained on these datasets limited as an assistive technology in real-world settings, such as helping visually impaired people navigate and accomplish everyday tasks. This gap motivated the introduction of the novel VizWiz dataset, which consists of images taken by the visually impaired and captions that have useful, task-oriented information. In an attempt to help the machine learning computer vision field realize its promise of producing technologies that have positive social impact, the curators of the VizWiz dataset host several competitions, including one for image captioning. This work details the theory and engineering from our winning submission to the 2020 captioning competition. Our work provides a step towards improved assistive image captioning systems.

(2 min) We show that the influence of a subset of the training samples can be removed -- or "forgotten" -- from the weights of a network trained on large-scale image classification tasks, and we provide strong computable bounds on the amount of remaining information after forgetting. Inspired by real-world applications of forgetting techniques, we introduce a novel notion of forgetting in mixed-privacy setting, where we know that a "core" subset of the training samples does not need to be forgotten. While this variation of the problem is conceptually simple, we show that working in this setting significantly improves the accuracy and guarantees of forgetting methods applied to vision classification tasks. Moreover, our method allows efficient removal of all information contained in non-core data by simply setting to zero a subset of the weights with minimal loss in performance. We achieve these results by replacing a standard deep network with a suitable linear approximation. With opportune changes to the network architecture and training procedure, we show that such linear approximation achieves comparable performance to the original network and that the forgetting problem becomes quadratic and can be solved efficiently even for large models. Unlike previous forgetting methods on deep networks, ours can achieve close to the state-of-the-art accuracy on large scale vision tasks. In particular, we show that our method allows forgetting without having to trade off the model accuracy.

(2 min) The phenomenon of adversarial examples illustrates one of the most basic vulnerabilities of deep neural networks. Among the variety of techniques introduced to surmount this inherent weakness, adversarial training has emerged as the most common and efficient strategy to achieve robustness. Typically, this is achieved by balancing robust and natural objectives. In this work, we aim to achieve better trade-off between robust and natural performances by enforcing a domain-invariant feature representation. We present a new adversarial training method, Domain Invariant Adversarial Learning (DIAL), which learns a feature representation which is both robust and domain invariant. DIAL uses a variant of Domain Adversarial Neural Network (DANN) on the natural domain and its corresponding adversarial domain. In a case where the source domain consists of natural examples and the target domain is the adversarially perturbed examples, our method learns a feature representation constrained not to discriminate between the natural and adversarial examples, and can therefore achieve a more robust representation. Our experiments indicate that our method improves both robustness and natural accuracy, when compared to current state-of-the-art adversarial training methods.

(2 min) Topological Data Analysis (TDA) has emerged recently as a robust tool to extract and compare the structure of datasets. TDA identifies features in data such as connected components and holes and assigns a quantitative measure to these features. Several studies reported that topological features extracted by TDA tools provide unique information about the data, discover new insights, and determine which feature is more related to the outcome. On the other hand, the overwhelming success of deep neural networks in learning patterns and relationships has been proven on a vast array of data applications, images in particular. To capture the characteristics of both powerful tools, we propose \textit{TDA-Net}, a novel ensemble network that fuses topological and deep features for the purpose of enhancing model generalizability and accuracy. We apply the proposed \textit{TDA-Net} to a critical application, which is the automated detection of COVID-19 from CXR images. The experimental results showed that the proposed network achieved excellent performance and suggests the applicability of our method in practice.

(2 min) There has recently been significant interest in training reinforcement learning (RL) agents in vision-based environments. This poses many challenges, such as high dimensionality and potential for observational overfitting through spurious correlations. A promising approach to solve both of these problems is a self-attention bottleneck, which provides a simple and effective framework for learning high performing policies, even in the presence of distractions. However, due to poor scalability of attention architectures, these methods do not scale beyond low resolution visual inputs, using large patches (thus small attention matrices). In this paper we make use of new efficient attention algorithms, recently shown to be highly effective for Transformers, and demonstrate that these new techniques can be applied in the RL setting. This allows our attention-based controllers to scale to larger visual inputs, and facilitate the use of smaller patches, even individual pixels, improving generalization. In addition, we propose a new efficient algorithm approximating softmax attention with what we call hybrid random features, leveraging the theory of angular kernels. We show theoretically and empirically that hybrid random features is a promising approach when using attention for vision-based RL.

(2 min) Recently spiking neural networks (SNNs), the third-generation of neural networks has shown remarkable capabilities of energy-efficient computing, which is a promising alternative for deep neural networks (DNNs) with high energy consumption. SNNs have reached competitive results compared to DNNs in relatively simple tasks and small datasets such as image classification and MNIST/CIFAR, while few studies on more challenging vision tasks on complex datasets. In this paper, we focus on extending deep SNNs to object tracking, a more advanced vision task with embedded applications and energy-saving requirements, and present a spike-based Siamese network called SiamSNN. Specifically, we propose an optimized hybrid similarity estimation method to exploit temporal information in the SNNs, and introduce a novel two-status coding scheme to optimize the temporal distribution of output spike trains for further improvements. SiamSNN is the first deep SNN tracker that achieves short latency and low precision loss on the visual object tracking benchmarks OTB2013/2015, VOT2016/2018, and GOT-10k. Moreover, SiamSNN achieves notably low energy consumption and real-time on Neuromorphic chip TrueNorth.

(2 min) Background and Objective: Manually annotating gastric X-ray images for gastritis detection is time-consuming and expensive because it typically requires expert knowledge. This paper proposes a self-supervised learning method to solve this problem. This study aims to verify the effectiveness of the proposed self-supervised learning method in gastritis detection using a few annotated gastric X-ray images. Methods: In this paper, we propose a novel self-supervised learning method that can perform explicit self-supervised learning and learn discriminative representations from gastric X-ray images. Models trained with the proposed method were fine-tuned on datasets with a few annotated gastric X-ray images. For comparison, several state-of-the-art self-supervised learning methods, i.e., containing SimSiam, BYOL, PIRL-jigsaw, PIRL-rotation, and SimCLR, were compared with the proposed method. Furthermore, two baseline methods, one pretrained on ImageNet and the other trained from scratch, were compared with the proposed method. Results: The proposed method's harmonic mean score of sensitivity and specificity after fine-tuning with the annotated data of 10, 20, 30, and 40 patients were 0.875, 0.911, 0.915, and 0.931, respectively. The proposed method outperformed all comparative methods, including the five state-of-the-art self-supervised learning and two baseline methods. Experimental results showed the effectiveness of the proposed method in gastritis detection with a few annotated gastric X-ray images. Conclusions: The proposed self-supervised learning method shows potential for clinical use in gastritis detection using a few annotated gastric X-ray images.

(2 min) This paper serves as a survey of recent advances in large margin training and its theoretical foundations, mostly for (nonlinear) deep neural networks (DNNs) that are probably the most prominent machine learning models for large-scale data in the community over the past decade. We generalize the formulation of classification margins from classical research to latest DNNs, summarize theoretical connections between the margin, network generalization, and robustness, and introduce recent efforts in enlarging the margins for DNNs comprehensively. Since the viewpoint of different methods is discrepant, we categorize them into groups for ease of comparison and discussion in the paper. Hopefully, our discussions and overview inspire new research work in the community that aim to improve the performance of DNNs, and we also point to directions where the large margin principle can be verified to provide theoretical evidence why certain regularizations for DNNs function well in practice. We managed to shorten the paper such that the crucial spirit of large margin learning and related methods are better emphasized.

(2 min) Glaucoma is the leading cause of irreversible blindness in the world, affecting over 70 million people. The cumbersome Standard Automated Perimetry (SAP) test is most frequently used to detect visual loss due to glaucoma. Due to the SAP test's innate difficulty and its high test-retest variability, we propose the RetiNerveNet, a deep convolutional recursive neural network for obtaining estimates of the SAP visual field. RetiNerveNet uses information from the more objective Spectral-Domain Optical Coherence Tomography (SDOCT). RetiNerveNet attempts to trace-back the arcuate convergence of the retinal nerve fibers, starting from the Retinal Nerve Fiber Layer (RNFL) thickness around the optic disc, to estimate individual age-corrected 24-2 SAP values. Recursive passes through the proposed network sequentially yield estimates of the visual locations progressively farther from the optic disc. While all the methods used for our experiments exhibit lower performance for the advanced disease group, the proposed network is observed to be more accurate than all the baselines for estimating the individual visual field values. We further augment RetiNerveNet to additionally predict the SAP Mean Deviation values and also create an ensemble of RetiNerveNets that further improves the performance, by increasingly weighting-up underrepresented parts of the training data.

(2 min) Facial action unit (AU) detection in the wild is a challenging problem, due to the unconstrained variability in facial appearances and the lack of accurate annotations. Most existing methods depend on either impractical labor-intensive labeling or inaccurate pseudo labels. In this paper, we propose an end-to-end unconstrained facial AU detection framework based on domain adaptation, which transfers accurate AU labels from a constrained source domain to an unconstrained target domain by exploiting labels of AU-related facial landmarks. Specifically, we map a source image with label and a target image without label into a latent feature domain by combining source landmark-related feature with target landmark-free feature. Due to the combination of source AU-related information and target AU-free information, the latent feature domain with transferred source label can be learned by maximizing the target-domain AU detection performance. Moreover, we introduce a novel landmark adversarial loss to disentangle the landmark-free feature from the landmark-related feature by treating the adversarial learning as a multi-player minimax game. Our framework can also be naturally extended for use with target-domain pseudo AU labels. Extensive experiments show that our method soundly outperforms lower-bounds and upper-bounds of the basic model, as well as state-of-the-art approaches on the challenging in-the-wild benchmarks. The code is available at https://github.com/ZhiwenShao/ADLD.

(2 min) Recently, some approaches are proposed to harness deep convolutional networks to facilitate superpixel segmentation. The common practice is to first evenly divide the image into a pre-defined number of grids and then learn to associate each pixel with its surrounding grids. However, simply applying a series of convolution operations with limited receptive fields can only implicitly perceive the relations between the pixel and its surrounding grids. Consequently, existing methods often fail to provide an effective context when inferring the association map. To remedy this issue, we propose a novel \textbf{A}ssociation \textbf{I}mplantation (AI) module to enable the network to explicitly capture the relations between the pixel and its surrounding grids. The proposed AI module directly implants the features of grid cells to the surrounding of its corresponding central pixel, and conducts convolution on the padded window to adaptively transfer knowledge between them. With such an implantation operation, the network could explicitly harvest the pixel-grid level context, which is more in line with the target of superpixel segmentation comparing to the pixel-wise relation. Furthermore, to pursue better boundary precision, we design a boundary-perceiving loss to help the network discriminate the pixels around boundaries in hidden feature level, which could benefit the subsequent inferring modules to accurately identify more boundary pixels. Extensive experiments on BSDS500 and NYUv2 datasets show that our method could not only achieve state-of-the-art performance but maintain satisfactory inference efficiency.

(2 min) This paper introduces the Diabetic Foot Ulcers dataset (DFUC2021) for analysis of pathology, focusing on infection and ischaemia. We describe the data preparation of DFUC2021 for ground truth annotation, data curation and data analysis. The final release of DFUC2021 consists of 15,683 DFU patches, with 5,955 training, 5,734 for testing and 3,994 unlabeled DFU patches. The ground truth labels are four classes, i.e. control, infection, ischaemia and both conditions. We curate the dataset using image hashing techniques and analyse the separability using UMAP projection. We benchmark the performance of five key backbones of deep learning, i.e. VGG16, ResNet101, InceptionV3, DenseNet121 and EfficientNet on DFUC2021. We report the optimised results of these key backbones with different strategies. Based on our observations, we conclude that EfficientNetB0 with data augmentation and transfer learning provided the best results for multi-class (4-class) classification with macro-average Precision, Recall and F1-score of 0.57, 0.62 and 0.55, respectively. In ischaemia and infection recognition, when trained on one-versus-all, EfficientNetB0 achieved comparable results with the state of the art. Finally, we interpret the results with statistical analysis and Grad-CAM visualisation.

(2 min) Few-shot dataset generalization is a challenging variant of the well-studied few-shot classification problem where a diverse training set of several datasets is given, for the purpose of training an adaptable model that can then learn classes from new datasets using only a few examples. To this end, we propose to utilize the diverse training set to construct a universal template: a partial model that can define a wide array of dataset-specialized models, by plugging in appropriate components. For each new few-shot classification problem, our approach therefore only requires inferring a small number of parameters to insert into the universal template. We design a separate network that produces an initialization of those parameters for each given task, and we then fine-tune its proposed initialization via a few steps of gradient descent. Our approach is more parameter-efficient, scalable and adaptable compared to previous methods, and achieves the state-of-the-art on the challenging Meta-Dataset benchmark.

(2 min) In the era of big data, methods for improving memory and computational efficiency have become crucial for successful deployment of technologies. Hashing is one of the most effective approaches to deal with computational limitations that come with big data. One natural way for formulating this problem is spectral hashing that directly incorporates affinity to learn binary codes. However, due to binary constraints, the optimization becomes intractable. To mitigate this challenge, different relaxation approaches have been proposed to reduce the computational load of obtaining binary codes and still attain a good solution. The problem with all existing relaxation methods is resorting to one or more additional auxiliary variables to attain high quality binary codes while relaxing the problem. The existence of auxiliary variables leads to coordinate descent approach which increases the computational complexity. We argue that introducing these variables is unnecessary. To this end, we propose a novel relaxed formulation for spectral hashing that adds no additional variables to the problem. Furthermore, instead of solving the problem in original space where number of variables is equal to the data points, we solve the problem in a much smaller space and retrieve the binary codes from this solution. This trick reduces both the memory and computational complexity at the same time. We apply two optimization techniques, namely projected gradient and optimization on manifold, to obtain the solution. Using comprehensive experiments on four public datasets, we show that the proposed efficient spectral hashing (ESH) algorithm achieves highly competitive retrieval performance compared with state of the art at low complexity.

(2 min) Deep Neural Networks (DNNs) have achieved remarkable success in many computer vision tasks recently, but the huge number of parameters and the high computation overhead hinder their deployments on resource-constrained edge devices. It is worth noting that channel pruning is an effective approach for compressing DNN models. A critical challenge is to determine which channels are to be removed, so that the model accuracy will not be negatively affected. In this paper, we first propose Spatial and Channel Attention (SCA), a new attention module combining both spatial and channel attention that respectively focuses on "where" and "what" are the most informative parts. Guided by the scale values generated by SCA for measuring channel importance, we further propose a new channel pruning approach called Channel Pruning guided by Spatial and Channel Attention (CPSCA). Experimental results indicate that SCA achieves the best inference accuracy, while incurring negligibly extra resource consumption, compared to other state-of-the-art attention modules. Our evaluation on two benchmark datasets shows that, with the guidance of SCA, our CPSCA approach achieves higher inference accuracy than other state-of-the-art pruning methods under the same pruning ratios.

(2 min) Early and reliable COVID-19 diagnosis based on chest 3-D CT scans can assist medical specialists in vital circumstances. Deep learning methodologies constitute a main approach for chest CT scan analysis and disease prediction. However, large annotated databases are necessary for developing deep learning models that are able to provide COVID-19 diagnosis across various medical environments in different countries. Due to privacy issues, publicly available COVID-19 CT datasets are highly difficult to obtain, which hinders the research and development of AI-enabled diagnosis methods of COVID-19 based on CT scans. In this paper we present the COV19-CT-DB database which is annotated for COVID-19, consisting of about 5,000 3-D CT scans, We have split the database in training, validation and test datasets. The former two datasets can be used for training and validation of machine learning models, while the latter will be used for evaluation of the developed models. We also present a deep learning approach, based on a CNN-RNN network and report its performance on the COVID19-CT-DB database.

(2 min) Recent advances in document image analysis (DIA) have been primarily driven by the application of neural networks. Ideally, research outcomes could be easily deployed in production and extended for further investigation. However, various factors like loosely organized codebases and sophisticated model configurations complicate the easy reuse of important innovations by a wide audience. Though there have been on-going efforts to improve reusability and simplify deep learning (DL) model development in disciplines like natural language processing and computer vision, none of them are optimized for challenges in the domain of DIA. This represents a major gap in the existing toolkit, as DIA is central to academic research across a wide range of disciplines in the social sciences and humanities. This paper introduces layoutparser, an open-source library for streamlining the usage of DL in DIA research and applications. The core layoutparser library comes with a set of simple and intuitive interfaces for applying and customizing DL models for layout detection, character recognition, and many other document processing tasks. To promote extensibility, layoutparser also incorporates a community platform for sharing both pre-trained models and full document digitization pipelines. We demonstrate that layoutparser is helpful for both lightweight and large-scale digitization pipelines in real-word use cases. The library is publicly available at https://layout-parser.github.io/.

(2 min) A video prediction model that generalizes to diverse scenes would enable intelligent agents such as robots to perform a variety of tasks via planning with the model. However, while existing video prediction models have produced promising results on small datasets, they suffer from severe underfitting when trained on large and diverse datasets. To address this underfitting challenge, we first observe that the ability to train larger video prediction models is often bottlenecked by the memory constraints of GPUs or TPUs. In parallel, deep hierarchical latent variable models can produce higher quality predictions by capturing the multi-level stochasticity of future observations, but end-to-end optimization of such models is notably difficult. Our key insight is that greedy and modular optimization of hierarchical autoencoders can simultaneously address both the memory constraints and the optimization challenges of large-scale video prediction. We introduce Greedy Hierarchical Variational Autoencoders (GHVAEs), a method that learns high-fidelity video predictions by greedily training each level of a hierarchical autoencoder. In comparison to state-of-the-art models, GHVAEs provide 17-55% gains in prediction performance on four video datasets, a 35-40% higher success rate on real robot tasks, and can improve performance monotonically by simply adding more modules.

(2 min) Routine and consistent data collection is required to address contemporary transportation issues.The cost of data collection increases significantly when sophisticated machines are used to collect data. Due to this constraint, State Departments of Transportation struggles to collect consistent data for analyzing and resolving transportation problems in a timely manner. Recent advancements in the sensors integrated into smartphones have resulted in a more affordable method of data collection.The primary objective of this study is to develop and implement a smartphone application for data collection.The currently designed app consists of three major modules: a frontend graphical user interface (GUI), a sensor module, and a backend module. While the frontend user interface enables interaction with the app, the sensor modules collect relevant data such as video and accelerometer readings while the app is in use. The backend, on the other hand, is made up of firebase storage, which is used to store the gathered data.In comparison to other developed apps for collecting pavement information, this current app is not overly reliant on the internet enabling the app to be used in areas of restricted internet access.The developed application was evaluated by collecting data on the i70W highway connecting Columbia, Missouri, and Kansas City, Missouri.The data was analyzed for a variety of purposes, including calculating the International Roughness Index (IRI), identifying pavement distresses, and understanding driver's behaviour and environment .The results of the application indicate that the data collected by the app is of high quality.

(2 min) This paper provides a novel approach to stitching surface images of rotationally symmetric parts. It presents a process pipeline that uses a feature-based stitching approach to create a distortion-free and true-to-life image from a video file. The developed process thus enables, for example, condition monitoring without having to view many individual images. For validation purposes, this will be demonstrated in the paper using the concrete example of a worn ball screw drive spindle. The developed algorithm aims at reproducing the functional principle of a line scan camera system, whereby the physical measuring systems are replaced by a feature-based approach. For evaluation of the stitching algorithms, metrics are used, some of which have only been developed in this work or have been supplemented by test procedures already in use. The applicability of the developed algorithm is not only limited to machine tool spindles. Instead, the developed method allows a general approach to the surface inspection of various rotationally symmetric components and can therefore be used in a variety of industrial applications. Deep-learning-based detection Algorithms can easily be implemented to generate a complete pipeline for failure detection and condition monitoring on rotationally symmetric parts.

(2 min) General segmentation models downsample images and then upsample to restore resolution for pixel level prediction. In such schema, upsample technique is vital in maintaining information for better performance. In this paper, we present a new upsample approach, Attention Upsample (AU), that could serve as general upsample method and be incorporated into any segmentation model that possesses lateral connections. AU leverages pixel-level attention to model long range dependency and global information for better reconstruction. It consists of Attention Decoder (AD) and bilinear upsample as residual connection to complement the upsampled features. AD adopts the idea of decoder from transformer which upsamples features conditioned on local and detailed information from contracting path. Moreover, considering the extensive memory and computation cost of pixel-level attention, we further propose to use window attention scheme to restrict attention computation in local windows instead of global range. Incorporating window attention, we denote our decoder as Window Attention Decoder (WAD) and our upsample method as Window Attention Upsample (WAU). We test our method on classic U-Net structure with lateral connection to deliver information from contracting path and achieve state-of-the-arts performance on Synapse (80.30 DSC and 23.12 HD) and MSD Brain (74.75 DSC) datasets.

(2 min) GAN inversion aims to invert a given image back into the latent space of a pretrained GAN model, for the image to be faithfully reconstructed from the inverted code by the generator. As an emerging technique to bridge the real and fake image domains, GAN inversion plays an essential role in enabling the pretrained GAN models such as StyleGAN and BigGAN to be used for real image editing applications. Meanwhile, GAN inversion also provides insights on the interpretation of GAN's latent space and how the realistic images can be generated. In this paper, we provide an overview of GAN inversion with a focus on its recent algorithms and applications. We cover important techniques of GAN inversion and their applications to image restoration and image manipulation. We further elaborate on some trends and challenges for future directions.

(2 min) Contrastive loss and its variants have become very popular recently for learning visual representations without supervision. In this work, we study three intriguing properties of contrastive learning. We first generalize the standard contrastive loss to a broader family of losses, and we find that various instantiations of the generalized loss perform similarly under the presence of a multi-layer non-linear projection head. We then study if instance-based contrastive learning (such as in SimCLR, MoCo, BYOL, and so on, which are based on global image representation) can learn well on images with multiple objects present. We find that meaningful hierarchical local features can be learned despite the fact that these objectives operate on global instance-level features. Finally, we study an intriguing phenomenon of feature suppression among competing features shared across augmented views, such as "color distribution" vs "object class". We construct datasets with explicit and controllable competing features, and show that, for contrastive learning, a few bits of easy-to-learn shared features can suppress, and even fully prevent, the learning of other sets of competing features. In scenarios where there are multiple objects in an image, the dominant object would suppress the learning of smaller objects. Existing contrastive learning methods critically rely on data augmentation to favor certain sets of features over others, and face potential limitation for scenarios where existing augmentations cannot fully address the feature suppression. This poses open challenges to existing contrastive learning techniques.

(2 min) Stereotactic body radiation therapy allows for a precise and accurate dose delivery. Organ motion during treatment bears the risk of undetected high dose healthy tissue exposure. An organ very susceptible to high dose is the oesophagus. Its low contrast on CT and the oblong shape renders motion estimation difficult. We tackle this issue by modern algorithms to measure the oesophageal motion voxel-wise and to estimate motion related dosimetric impact. Oesophageal motion was measured using deformable image registration and 4DCT of 11 internal and 5 public datasets. Current clinical practice of contouring the organ on 3DCT was compared to timely resolved 4DCT contours. The dosimetric impact of the motion was estimated by analysing the trajectory of each voxel in the 4D dose distribution. Finally an organ motion model was built, allowing for easier patient-wise comparisons. Motion analysis showed mean absolute maximal motion amplitudes of 4.55 +/- 1.81 mm left-right, 5.29 +/- 2.67 mm anterior-posterior and 10.78 +/- 5.30 mm superior-inferior. Motion between the cohorts differed significantly. In around 50 % of the cases the dosimetric passing criteria was violated. Contours created on 3DCT did not cover 14 % of the organ for 50 % of the respiratory cycle and the 3D contour is around 38 % smaller than the union of all 4D contours. The motion model revealed that the maximal motion is not limited to the lower part of the organ. Our results showed motion amplitudes higher than most reported values in the literature and that motion is very heterogeneous across patients. Therefore, individual motion information should be considered in contouring and planning.

(2 min) The usage of environment sensor models for virtual testing is a promising approach to reduce the testing effort of autonomous driving. However, in order to deduce any statements regarding the performance of an autonomous driving function based on simulation, the sensor model has to be validated to determine the discrepancy between the synthetic and real sensor data. Since a certain degree of divergence can be assumed to exist, the sufficient level of fidelity must be determined, which poses a major challenge. In particular, a method for quantifying the fidelity of a sensor model does not exist and the problem of defining an appropriate metric remains. In this work, we train a neural network to distinguish real and simulated radar sensor data with the purpose of learning the latent features of real radar point clouds. Furthermore, we propose the classifier's confidence score for the `real radar point cloud' class as a metric to determine the degree of fidelity of synthetically generated radar data. The presented approach is evaluated and it can be demonstrated that the proposed deep evaluation metric outperforms conventional metrics in terms of its capability to identify characteristic differences between real and simulated radar data.

(2 min) Scene graph generation (SGG) is to detect entity pairs with their relations in an image. Existing SGG approaches often use multi-stage pipelines to decompose this task into object detection, relation graph construction, and dense or dense-to-sparse relation prediction. Instead, from a perspective on SGG as a direct set prediction, this paper presents a simple, sparse, and unified framework for relation detection, termed as Structured Sparse R-CNN. The key to our method is a set of learnable triplet queries and structured triplet detectors which could be jointly optimized from the training set in an end-to-end manner. Specifically, the triplet queries encode the general prior for entity pair locations, categories, and their relations, and provide an initial guess of relation detection for subsequent refinement. The triplet detector presents a cascaded dynamic head design to progressively refine the results of relation detection. In addition, to relieve the training difficulty of Structured Sparse R-CNN, we propose a relaxed and enhanced training strategy based on knowledge distillation from a Siamese Sparse R-CNN. We also propose adaptive focusing parameter and average logit approach for imbalance data distribution. We perform experiments on two benchmarks: Visual Genome and Open Images, and the results demonstrate that our method achieves the state-of-the-art performance. Meanwhile, we perform in-depth ablation studies to provide insights on our structured modeling in triplet detector design and training strategies.

(2 min) This is a technical report for CVPR 2021 AliProducts Challenge. AliProducts Challenge is a competition proposed for studying the large-scale and fine-grained commodity image recognition problem encountered by worldleading ecommerce companies. The large-scale product recognition simultaneously meets the challenge of noisy annotations, imbalanced (long-tailed) data distribution and fine-grained classification. In our solution, we adopt stateof-the-art model architectures of both CNNs and Transformer, including ResNeSt, EfficientNetV2, and DeiT. We found that iterative data cleaning, classifier weight normalization, high-resolution finetuning, and test time augmentation are key components to improve the performance of training with the noisy and imbalanced dataset. Finally, we obtain 6.4365% mean class error rate in the leaderboard with our ensemble model.

(2 min) Training images with data transformations have been suggested as contrastive examples to complement the testing set for generalization performance evaluation of deep neural networks (DNNs). In this work, we propose a practical framework ContRE (The word "contre" means "against" or "versus" in French.) that uses Contrastive examples for DNN geneRalization performance Estimation. Specifically, ContRE follows the assumption in contrastive learning that robust DNN models with good generalization performance are capable of extracting a consistent set of features and making consistent predictions from the same image under varying data transformations. Incorporating with a set of randomized strategies for well-designed data transformations over the training set, ContRE adopts classification errors and Fisher ratios on the generated contrastive examples to assess and analyze the generalization performance of deep models in complement with a testing set. To show the effectiveness and the efficiency of ContRE, extensive experiments have been done using various DNN models on three open source benchmark datasets with thorough ablation studies and applicability analyses. Our experiment results confirm that (1) behaviors of deep models on contrastive examples are strongly correlated to what on the testing set, and (2) ContRE is a robust measure of generalization performance complementing to the testing set in various settings.

(2 min) Random field and random cluster theory are used to prove certain mathematical results concerning the probability distribution of image pixel intensities characterized as generic $2D$ integer arrays. The size of the smallest bounded region within an image is estimated for segmenting an image, from which, the equilibrium distribution of intensities can be recovered. From the estimated bounded regions, properties of the sub-optimal and equilibrium distributions of intensities are derived, which leads to an image compression methodology whereby only slightly more than half of all pixels are required for a worst-case reconstruction of the original image. An example in unsupervised object detection illustrates the mathematical results.

(2 min) This paper presents a deep learning-based point cloud processing method named FloorPP-Net for the task of Scan-to-BIM (building information model). FloorPP-Net first converts the input point cloud of a building story into point pillars (PP), then predicts the corners and edges to output the floor plan. Altogether, FloorPP-Net establishes an end-to-end supervised learning framework for the Scan-to-Floor-Plan (Scan2FP) task. In the 1st International Scan-to-BIM Challenge held in conjunction with CVPR 2021, FloorPP-Net was ranked the second runner-up in the floor plan reconstruction track. Future work includes general edge proposals, 2D plan regularization, and 3D BIM reconstruction.

(2 min) Feature learning for 3D object detection from point clouds is very challenging due to the irregularity of 3D point cloud data. In this paper, we propose Pointformer, a Transformer backbone designed for 3D point clouds to learn features effectively. Specifically, a Local Transformer module is employed to model interactions among points in a local region, which learns context-dependent region features at an object level. A Global Transformer is designed to learn context-aware representations at the scene level. To further capture the dependencies among multi-scale representations, we propose Local-Global Transformer to integrate local features with global features from higher resolution. In addition, we introduce an efficient coordinate refinement module to shift down-sampled points closer to object centroids, which improves object proposal generation. We use Pointformer as the backbone for state-of-the-art object detection models and demonstrate significant improvements over original models on both indoor and outdoor datasets.

(2 min) In this paper, we present a novel adversarial lossy video compression model. At extremely low bit-rates, standard video coding schemes suffer from unpleasant reconstruction artifacts such as blocking, ringing etc. Existing learned neural approaches to video compression have achieved reasonable success on reducing the bit-rate for efficient transmission and reduce the impact of artifacts to an extent. However, they still tend to produce blurred results under extreme compression. In this paper, we present a deep adversarial learned video compression model that minimizes an auxiliary adversarial distortion objective. We find this adversarial objective to correlate better with human perceptual quality judgement relative to traditional quality metrics such as MS-SSIM and PSNR. Our experiments using a state-of-the-art learned video compression system demonstrate a reduction of perceptual artifacts and reconstruction of detail lost especially under extremely high compression.

(2 min) Underwater image restoration is of significant importance in unveiling the underwater world. Numerous techniques and algorithms have been developed in the past decades. However, due to fundamental difficulties associated with imaging/sensing, lighting, and refractive geometric distortions, in capturing clear underwater images, no comprehensive evaluations have been conducted of underwater image restoration. To address this gap, we have constructed a large-scale real underwater image dataset, dubbed `HICRD' (Heron Island Coral Reef Dataset), for the purpose of benchmarking existing methods and supporting the development of new deep-learning based methods. We employ accurate water parameter (diffuse attenuation coefficient) in generating reference images. There are 2000 reference restored images and 6003 original underwater images in the unpaired training set. Further, we present a novel method for underwater image restoration based on unsupervised image-to-image translation framework. Our proposed method leveraged contrastive learning and generative adversarial networks to maximize the mutual information between raw and restored images. Extensive experiments with comparisons to recent approaches further demonstrate the superiority of our proposed method. Our code and dataset are publicly available at GitHub.

(2 min) Deep visual models are susceptible to adversarial perturbations to inputs. Although these signals are carefully crafted, they still appear noise-like patterns to humans. This observation has led to the argument that deep visual representation is misaligned with human perception. We counter-argue by providing evidence of human-meaningful patterns in adversarial perturbations. We first propose an attack that fools a network to confuse a whole category of objects (source class) with a target label. Our attack also limits the unintended fooling by samples from non-sources classes, thereby circumscribing human-defined semantic notions for network fooling. We show that the proposed attack not only leads to the emergence of regular geometric patterns in the perturbations, but also reveals insightful information about the decision boundaries of deep models. Exploring this phenomenon further, we alter the `adversarial' objective of our attack to use it as a tool to `explain' deep visual representation. We show that by careful channeling and projection of the perturbations computed by our method, we can visualize a model's understanding of human-defined semantic notions. Finally, we exploit the explanability properties of our perturbations to perform image generation, inpainting and interactive image manipulation by attacking adversarialy robust `classifiers'.In all, our major contribution is a novel pragmatic adversarial attack that is subsequently transformed into a tool to interpret the visual models. The article also makes secondary contributions in terms of establishing the utility of our attack beyond the adversarial objective with multiple interesting applications.

(2 min) Can a computer determine a piano player's skill level? Is it preferable to base this assessment on visual analysis of the player's performance or should we trust our ears over our eyes? Since current CNNs have difficulty processing long video videos, how can shorter clips be sampled to best reflect the players skill level? In this work, we collect and release a first-of-its-kind dataset for multimodal skill assessment focusing on assessing piano player's skill level, answer the asked questions, initiate work in automated evaluation of piano playing skills and provide baselines for future work. Dataset is available from: https://github.com/ParitoshParmar/Piano-Skills-Assessment.

(2 min) In this paper, we present a novel approach to the audio-visual video parsing (AVVP) task that demarcates events from a video separately for audio and visual modalities. The proposed parsing approach simultaneously detects the temporal boundaries in terms of start and end times of such events. We show how AVVP can benefit from the following techniques geared towards effective cross-modal learning: (i) adversarial training and skip connections (ii) global context aware attention and, (iii) self-supervised pretraining using an audio-video grounding objective to obtain cross-modal audio-video representations. We present extensive experimental evaluations on the Look, Listen, and Parse (LLP) dataset and show that we outperform the state-of-the-art Hybrid Attention Network (HAN) on all five metrics proposed for AVVP. We also present several ablations to validate the effect of pretraining, global attention and adversarial training.

(2 min) We propose an effective two-stage approach to tackle the problem of language-based Human-centric Spatio-Temporal Video Grounding (HC-STVG) task. In the first stage, we propose an Augmented 2D Temporal Adjacent Network (Augmented 2D-TAN) to temporally ground the target moment corresponding to the given description. Primarily, we improve the original 2D-TAN from two aspects: First, a temporal context-aware Bi-LSTM Aggregation Module is developed to aggregate clip-level representations, replacing the original max-pooling. Second, we propose to employ Random Concatenation Augmentation (RCA) mechanism during the training phase. In the second stage, we use pretrained MDETR model to generate per-frame bounding boxes via language query, and design a set of hand-crafted rules to select the best matching bounding box outputted by MDETR for each frame within the grounded moment.

(2 min) Few-shot object detection (FSOD) aims to detect objects using only few examples. It's critically needed for many practical applications but so far remains challenging. We propose a meta-learning based few-shot object detection method by transferring meta-knowledge learned from data-abundant base classes to data-scarce novel classes. Our method incorporates a coarse-to-fine approach into the proposal based object detection framework and integrates prototype based classifiers into both the proposal generation and classification stages. To improve proposal generation for few-shot novel classes, we propose to learn a lightweight matching network to measure the similarity between each spatial position in the query image feature map and spatially-pooled class features, instead of the traditional object/nonobject classifier, thus generating category-specific proposals and improving proposal recall for novel classes. To address the spatial misalignment between generated proposals and few-shot class examples, we propose a novel attentive feature alignment method, thus improving the performance of few-shot object detection. Meanwhile we jointly learn a Faster R-CNN detection head for base classes. Extensive experiments conducted on multiple FSOD benchmarks show our proposed approach achieves state of the art results under (incremental) few-shot learning settings.

(2 min) Post-hoc gradient-based interpretability methods [Simonyan et al., 2013, Smilkov et al., 2017] that provide instance-specific explanations of model predictions are often based on assumption (A): magnitude of input gradients -- gradients of logits with respect to input -- noisily highlight discriminative task-relevant features. In this work, we test the validity of assumption (A) using a three-pronged approach. First, we develop an evaluation framework, DiffROAR, to test assumption (A) on four image classification benchmarks. Our results suggest that (i) input gradients of standard models (i.e., trained on original data) may grossly violate (A), whereas (ii) input gradients of adversarially robust models satisfy (A). Second, we then introduce BlockMNIST, an MNIST-based semi-real dataset, that by design encodes a priori knowledge of discriminative features. Our analysis on BlockMNIST leverages this information to validate as well as characterize differences between input gradient attributions of standard and robust models. Finally, we theoretically prove that our empirical findings hold on a simplified version of the BlockMNIST dataset. Specifically, we prove that input gradients of standard one-hidden-layer MLPs trained on this dataset do not highlight instance-specific signal coordinates, thus grossly violating assumption (A). Our findings motivate the need to formalize and test common assumptions in interpretability in a falsifiable manner [Leavitt and Morcos, 2020]. Additionally, we believe that the DiffROAR evaluation framework and BlockMNIST-based datasets can serve as sanity checks to audit instance-specific interpretability methods.

(2 min) Computer vision techniques have attracted a great interest in precision agriculture, recently. The common goal of all computer vision-based precision agriculture tasks is to detect the objects of interest (e.g., crop, weed) and discriminating them from the background. The Weeds are unwanted plants growing among crops competing for nutrients, water, and sunlight, causing losses to crop yields. Weed detection and mapping is critical for site-specific weed management to reduce the cost of labor and impact of herbicides. This paper investigates the use of color and texture features for discrimination of Soybean crops and weeds. Feature extraction methods including two color spaces (RGB, HSV), gray level Co-occurrence matrix (GLCM), and Local Binary Pattern (LBP) are used to train the Support Vector Machine (SVM) classifier. The experiment was carried out on image dataset of soybean crop, obtained from an unmanned aerial vehicle (UAV), which is publicly available. The results from the experiment showed that the highest accuracy (above 96%) was obtained from the combination of color and LBP features.

(2 min) When dealing with deep neural network (DNN) applications on edge devices, continuously updating the model is important. Although updating a model with real incoming data is ideal, using all of them is not always feasible due to limits, such as labeling and communication costs. Thus, it is necessary to filter and select the data to use for training (i.e., active learning) on the device. In this paper, we formalize a practical active learning problem for DNNs on edge devices and propose a general task-agnostic framework to tackle this problem, which reduces it to a stream submodular maximization. This framework is light enough to be run with low computational resources, yet provides solutions whose quality is theoretically guaranteed thanks to the submodular property. Through this framework, we can configure data selection criteria flexibly, including using methods proposed in previous active learning studies. We evaluate our approach on both classification and object detection tasks in a practical setting to simulate a real-life scenario. The results of our study show that the proposed framework outperforms all other methods in both tasks, while running at a practical speed on real devices.

(2 min) A table arranging data in rows and columns is a very effective data structure, which has been widely used in business and scientific research. Considering large-scale tabular data in online and offline documents, automatic table recognition has attracted increasing attention from the document analysis community. Though human can easily understand the structure of tables, it remains a challenge for machines to understand that, especially due to a variety of different table layouts and styles. Existing methods usually model a table as either the markup sequence or the adjacency matrix between different table cells, failing to address the importance of the logical location of table cells, e.g., a cell is located in the first row and the second column of the table. In this paper, we reformulate the problem of table structure recognition as the table graph reconstruction, and propose an end-to-end trainable table graph reconstruction network (TGRNet) for table structure recognition. Specifically, the proposed method has two main branches, a cell detection branch and a cell logical location branch, to jointly predict the spatial location and the logical location of different cells. Experimental results on three popular table recognition datasets and a new dataset with table graph annotations (TableGraph-350K) demonstrate the effectiveness of the proposed TGRNet for table structure recognition. Code and annotations will be made publicly available.

(2 min) This paper reviews the current state of the art in Artificial Intelligence (AI) technologies and applications in the context of the creative industries. A brief background of AI, and specifically Machine Learning (ML) algorithms, is provided including Convolutional Neural Network (CNNs), Generative Adversarial Networks (GANs), Recurrent Neural Networks (RNNs) and Deep Reinforcement Learning (DRL). We categorise creative applications into five groups related to how AI technologies are used: i) content creation, ii) information analysis, iii) content enhancement and post production workflows, iv) information extraction and enhancement, and v) data compression. We critically examine the successes and limitations of this rapidly advancing technology in each of these areas. We further differentiate between the use of AI as a creative tool and its potential as a creator in its own right. We foresee that, in the near future, machine learning-based AI will be adopted widely as a tool or collaborative assistant for creativity. In contrast, we observe that the successes of machine learning in domains with fewer constraints, where AI is the `creator', remain modest. The potential of AI (or its developers) to win awards for its original creations in competition with human creatives is also limited, based on contemporary technologies. We therefore conclude that, in the context of creative industries, maximum benefit from AI will be derived where its focus is human centric -- where it is designed to augment, rather than replace, human creativity.

(2 min) Graphically-rich applications such as games are ubiquitous with attractive visual effects of Graphical User Interface (GUI) that offers a bridge between software applications and end-users. However, various types of graphical glitches may arise from such GUI complexity and have become one of the main component of software compatibility issues. Our study on bug reports from game development teams in NetEase Inc. indicates that graphical glitches frequently occur during the GUI rendering and severely degrade the quality of graphically-rich applications such as video games. Existing automated testing techniques for such applications focus mainly on generating various GUI test sequences and check whether the test sequences can cause crashes. These techniques require constant human attention to captures non-crashing bugs such as bugs causing graphical glitches. In this paper, we present the first step in automating the test oracle for detecting non-crashing bugs in graphically-rich applications. Specifically, we propose \texttt{GLIB} based on a code-based data augmentation technique to detect game GUI glitches. We perform an evaluation of \texttt{GLIB} on 20 real-world game apps (with bug reports available) and the result shows that \texttt{GLIB} can achieve 100\% precision and 99.5\% recall in detecting non-crashing bugs such as game GUI glitches. Practical application of \texttt{GLIB} on another 14 real-world games (without bug reports) further demonstrates that \texttt{GLIB} can effectively uncover GUI glitches, with 48 of 53 bugs reported by \texttt{GLIB} having been confirmed and fixed so far.

(2 min) One of the important and tedious task in agricultural practices is the detection of the disease on crops. It requires huge time as well as skilled labor. This paper proposes a smart and efficient technique for detection of crop disease which uses computer vision and machine learning techniques. The proposed system is able to detect 20 different diseases of 5 common plants with 93% accuracy.

(2 min) Despite recent progress of automatic medical image segmentation techniques, fully automatic results usually fail to meet the clinical use and typically require further refinement. In this work, we propose a quality-aware memory network for interactive segmentation of 3D medical images. Provided by user guidance on an arbitrary slice, an interaction network is firstly employed to obtain an initial 2D segmentation. The quality-aware memory network subsequently propagates the initial segmentation estimation bidirectionally over the entire volume. Subsequent refinement based on additional user guidance on other slices can be incorporated in the same manner. To further facilitate interactive segmentation, a quality assessment module is introduced to suggest the next slice to segment based on the current segmentation quality of each slice. The proposed network has two appealing characteristics: 1) The memory-augmented network offers the ability to quickly encode past segmentation information, which will be retrieved for the segmentation of other slices; 2) The quality assessment module enables the model to directly estimate the qualities of segmentation predictions, which allows an active learning paradigm where users preferentially label the lowest-quality slice for multi-round refinement. The proposed network leads to a robust interactive segmentation engine, which can generalize well to various types of user annotations (e.g., scribbles, boxes). Experimental results on various medical datasets demonstrate the superiority of our approach in comparison with existing techniques.

(2 min) A deraining network can be interpreted as a conditional generator that aims at removing rain streaks from image. Most existing image deraining methods ignore model errors caused by uncertainty that reduces embedding quality. Unlike existing image deraining methods that embed low-quality features into the model directly, we replace low-quality features by latent high-quality features. The spirit of closed-loop feedback in the automatic control field is borrowed to obtain latent high-quality features. A new method for error detection and feature compensation is proposed to address model errors. Extensive experiments on benchmark datasets as well as specific real datasets demonstrate that the proposed method outperforms recent state-of-the-art methods. Code is available at: \\ https://github.com/LI-Hao-SJTU/DerainRLNet

(2 min) Building facial analysis systems that generalize to extreme variations in lighting and facial expressions is a challenging problem that can potentially be alleviated using natural-looking synthetic data. Towards that, we propose LEGAN, a novel synthesis framework that leverages perceptual quality judgments for jointly manipulating lighting and expressions in face images, without requiring paired training data. LEGAN disentangles the lighting and expression subspaces and performs transformations in the feature space before upscaling to the desired output image. The fidelity of the synthetic image is further refined by integrating a perceptual quality estimation model, trained with face images rendered using multiple synthesis methods and their crowd-sourced naturalness ratings, into the LEGAN framework as an auxiliary discriminator. Using objective metrics like FID and LPIPS, LEGAN is shown to generate higher quality face images when compared with popular GAN models like StarGAN and StarGAN-v2 for lighting and expression synthesis. We also conduct a perceptual study using images synthesized by LEGAN and other GAN models and show the correlation between our quality estimation and visual fidelity. Finally, we demonstrate the effectiveness of LEGAN as training data augmenter for expression recognition and face verification tasks.

(2 min) Visual Question Answering system integrated with Unmanned Aerial Vehicle (UAV) has a lot of potentials to advance the post-disaster damage assessment purpose. Providing assistance to affected areas is highly dependent on real-time data assessment and analysis. Scope of the Visual Question Answering is to understand the scene and provide query related answer which certainly faster the recovery process after any disaster. In this work, we address the importance of \textit{visual question answering (VQA)} task for post-disaster damage assessment by presenting our recently developed VQA dataset called \textit{HurMic-VQA} collected during hurricane Michael, and comparing the performances of baseline VQA models.

(2 min) Backpropagation image saliency aims at explaining model predictions by estimating model-centric importance of individual pixels in the input. However, class-insensitivity of the earlier layers in a network only allows saliency computation with low resolution activation maps of the deeper layers, resulting in compromised image saliency. Remedifying this can lead to sanity failures. We propose CAMERAS, a technique to compute high-fidelity backpropagation saliency maps without requiring any external priors and preserving the map sanity. Our method systematically performs multi-scale accumulation and fusion of the activation maps and backpropagated gradients to compute precise saliency maps. From accurate image saliency to articulation of relative importance of input features for different models, and precise discrimination between model perception of visually similar objects, our high-resolution mapping offers multiple novel insights into the black-box deep visual models, which are presented in the paper. We also demonstrate the utility of our saliency maps in adversarial setup by drastically reducing the norm of attack signals by focusing them on the precise regions identified by our maps. Our method also inspires new evaluation metrics and a sanity check for this developing research direction. Code is available here https://github.com/VisMIL/CAMERAS

(2 min) We propose a semi-supervised approach for contemporary object detectors following the teacher-student dual model framework. Our method is featured with 1) the exponential moving averaging strategy to update the teacher from the student online, 2) using plenty of region proposals and soft pseudo-labels as the student's training targets, and 3) a light-weighted detection-specific data ensemble for the teacher to generate more reliable pseudo-labels. Compared to the recent state-of-the-art -- STAC, which uses hard labels on sparsely selected hard pseudo samples, the teacher in our model exposes richer information to the student with soft-labels on many proposals. Our model achieves COCO-style AP of 53.04% on VOC07 val set, 8.4% better than STAC, when using VOC12 as unlabeled data. On MS-COCO, it outperforms prior work when only a small percentage of data is taken as labeled. It also reaches 53.8% AP on MS-COCO test-dev with 3.1% gain over the fully supervised ResNet-152 Cascaded R-CNN, by tapping into unlabeled data of a similar size to the labeled data.

(2 min) Intelligent video summarization algorithms allow to quickly convey the most relevant information in videos through the identification of the most essential and explanatory content while removing redundant video frames. In this paper, we introduce the 3DST-UNet-RL framework for video summarization. A 3D spatio-temporal U-Net is used to efficiently encode spatio-temporal information of the input videos for downstream reinforcement learning (RL). An RL agent learns from spatio-temporal latent scores and predicts actions for keeping or rejecting a video frame in a video summary. We investigate if real/inflated 3D spatio-temporal CNN features are better suited to learn representations from videos than commonly used 2D image features. Our framework can operate in both, a fully unsupervised mode and a supervised training mode. We analyse the impact of prescribed summary lengths and show experimental evidence for the effectiveness of 3DST-UNet-RL on two commonly used general video summarization benchmarks. We also applied our method on a medical video summarization task. The proposed video summarization method has the potential to save storage costs of ultrasound screening videos as well as to increase efficiency when browsing patient video data during retrospective analysis or audit without loosing essential information

(2 min) Image compression using colour densities is historically impractical to decompress losslessly. We examine the use of conditional generative adversarial networks in making this transformation more feasible, through learning a mapping between the images and a loss function to train on. We show that this method is effective at producing visually lossless generations, indicating that efficient colour compression is viable.

(2 min) Transferability estimation is an essential problem in transfer learning to predict how good the performance is when transfer a source model (source task) to a target task. Recent analytical transferability metrics have been widely used for source model selection and multi-task learning. Earlier metrics does not work sufficiently well under the challenging cross-domain cross-task transfer settings, but recent OTCE score achieves a noteworthy performance using auxiliary tasks. A simplified version named OT-based NCE score sacrifices accuracy to be more efficient, but it can be further improved. Consequently, we propose a practical transferability metric called JC-NCE score to further improve the cross-domain cross-task transferability estimation performance, which is more efficient than the OTCE score and more accurate than the OT-based NCE score. Specifically, we build the joint correspondences between source and target data via solving an optimal transport problem with considering both the sample distance and label distance, and then compute the transferability score as the negative conditional entropy. Extensive validations under the intra-dataset and inter-dataset transfer settings demonstrate that our JC-NCE score outperforms the OT-based NCE score with about 7% and 12% gains, respectively.

(2 min) The COVID-19 pandemic started in China in December 2019 and quickly spread to several countries. The consequences of this pandemic are incalculable, causing the death of millions of people and damaging the global economy. To achieve large-scale control of this pandemic, fast tools for detection and treatment of patients are needed. Thus, the demand for alternative tools for the diagnosis of COVID-19 has increased dramatically since accurated and automated tools are not available. In this paper we present the ongoing work on a system for COVID-19 detection using ultrasound imaging and using Deep Learning techniques. Furthermore, such a system is implemented on a Raspberry Pi to make it portable and easy to use in remote regions without an Internet connection.

(2 min) In this paper, we explore contrastive learning for few-shot classification, in which we propose to use it as an additional auxiliary training objective acting as a data-dependent regularizer to promote more general and transferable features. In particular, we present a novel attention-based spatial contrastive objective to learn locally discriminative and class-agnostic features. As a result, our approach overcomes some of the limitations of the cross-entropy loss, such as its excessive discrimination towards seen classes, which reduces the transferability of features to unseen classes. With extensive experiments, we show that the proposed method outperforms state-of-the-art approaches, confirming the importance of learning good and transferable embeddings for few-shot learning.

(2 min) Occlusion is one of the most significant challenges encountered by object detectors and trackers. While both object detection and tracking has received a lot of attention in the past, most existing methods in this domain do not target detecting or tracking objects when they are occluded. However, being able to detect or track an object of interest through occlusion has been a long standing challenge for different autonomous tasks. Traditional methods that employ visual object trackers with explicit occlusion modeling experience drift and make several fundamental assumptions about the data. We propose to address this with a `tracking-by-detection` approach that builds upon the success of region based video object detectors. Our video level object detector uses a novel recurrent computational unit at its core that enables long term propagation of object features even under occlusion. Finally, we compare our approach with existing state-of-the-art video object detectors and show that our approach achieves superior results on a dataset of furniture assembly videos collected from the internet, where small objects like screws, nuts, and bolts often get occluded from the camera viewpoint.

(2 min) We present a novel neural surface reconstruction method, called NeuS, for reconstructing objects and scenes with high fidelity from 2D image inputs. Existing neural surface reconstruction approaches, such as DVR and IDR, require foreground mask as supervision, easily get trapped in local minima, and therefore struggle with the reconstruction of objects with severe self-occlusion or thin structures. Meanwhile, recent neural methods for novel view synthesis, such as NeRF and its variants, use volume rendering to produce a neural scene representation with robustness of optimization, even for highly complex objects. However, extracting high-quality surfaces from this learned implicit representation is difficult because there are not sufficient surface constraints in the representation. In NeuS, we propose to represent a surface as the zero-level set of a signed distance function (SDF) and develop a new volume rendering method to train a neural SDF representation. We observe that the conventional volume rendering method causes inherent geometric errors (i.e. bias) for surface reconstruction, and therefore propose a new formulation that is free of bias in the first order of approximation, thus leading to more accurate surface reconstruction even without the mask supervision. Experiments on the DTU dataset and the BlendedMVS dataset show that NeuS outperforms the state-of-the-arts in high-quality surface reconstruction, especially for objects and scenes with complex structures and self-occlusion.

(2 min) A new learning paradigm, self-supervised learning (SSL), can be used to solve such problems by pre-training a general model with large unlabeled images and then fine-tuning on a downstream task with very few labeled samples. Contrastive learning is a typical method of SSL, which can learn general invariant features. However, most of the existing contrastive learning is designed for classification tasks to obtain an image-level representation, which may be sub-optimal for semantic segmentation tasks requiring pixel-level discrimination. Therefore, we propose Global style and Local matching Contrastive Learning Network (GLCNet) for remote sensing semantic segmentation. Specifically, the global style contrastive module is used to learn an image-level representation better, as we consider the style features can better represent the overall image features; The local features matching contrastive module is designed to learn representations of local regions which is beneficial for semantic segmentation. We evaluate four remote sensing semantic segmentation datasets, and the experimental results show that our method mostly outperforms state-of-the-art self-supervised methods and ImageNet pre-training. Specifically, with 1\% annotation from the original dataset, our approach improves Kappa by 6\% on the ISPRS Potsdam dataset and 3\% on Deep Globe Land Cover Classification dataset relative to the existing baseline. Moreover, our method outperforms supervised learning when there are some differences between the datasets of upstream tasks and downstream tasks. Our study promotes the development of self-supervised learning in the field of remote sensing semantic segmentation. The source code is available at https://github.com/GeoX-Lab/G-RSIM.

(2 min) This technical report presents our solution to the HACS Temporal Action Localization Challenge 2021, Weakly-Supervised Learning Track. The goal of weakly-supervised temporal action localization is to temporally locate and classify action of interest in untrimmed videos given only video-level labels. We adopt the two-stream consensus network (TSCN) as the main framework in this challenge. The TSCN consists of a two-stream base model training procedure and a pseudo ground truth learning procedure. The base model training encourages the model to predict reliable predictions based on single modality (i.e., RGB or optical flow), based on the fusion of which a pseudo ground truth is generated and in turn used as supervision to train the base models. On the HACS v1.1.1 dataset, without fine-tuning the feature-extraction I3D models, our method achieves 22.20% on the validation set and 21.68% on the testing set in terms of average mAP. Our solution ranked the 2rd in this challenge, and we hope our method can serve as a baseline for future academic research.

(2 min) Metric learning is a widely used method for few shot learning in which the quality of prototypes plays a key role in the algorithm. In this paper we propose the trainable prototypes for distance measure instead of the artificial ones within the meta-training and task-training framework. Also to avoid the disadvantages that the episodic meta-training brought, we adopt non-episodic meta-training based on self-supervised learning. Overall we solve the few-shot tasks in two phases: meta-training a transferable feature extractor via self-supervised learning and training the prototypes for metric classification. In addition, the simple attention mechanism is used in both meta-training and task-training. Our method achieves state-of-the-art performance in a variety of established few-shot tasks on the standard few-shot visual classification dataset, with about 20% increase compared to the available unsupervised few-shot learning methods.

(2 min) It is widely accepted that reasoning about object shape is important for object recognition. However, the most powerful object recognition methods today do not explicitly make use of object shape during learning. In this work, motivated by recent developments in low-shot learning, findings in developmental psychology, and the increased use of synthetic data in computer vision research, we investigate how reasoning about 3D shape can be used to improve low-shot learning methods' generalization performance. We propose a new way to improve existing low-shot learning approaches by learning a discriminative embedding space using 3D object shape, and using this embedding by learning how to map images into it. Our new approach improves the performance of image-only low-shot learning approaches on multiple datasets. We also introduce Toys4K, a 3D object dataset with the largest number of object categories currently available, which supports low-shot learning.

(2 min) Many approaches to 3D image segmentation are based on hierarchical clustering of supervoxels into image regions. Here we describe a distributed algorithm capable of handling a tremendous number of supervoxels. The algorithm works recursively, the regions are divided into chunks that are processed independently in parallel by multiple workers. At each round of the recursive procedure, the chunk size in all dimensions are doubled until a single chunk encompasses the entire image. The final result is provably independent of the chunking scheme, and the same as if the entire image were processed without division into chunks. This is nontrivial because a pair of adjacent regions is scored by some statistical property (e.g. mean or median) of the affinities at the interface, and the interface may extend over arbitrarily many chunks. The trick is to delay merge decisions for regions that touch chunk boundaries, and only complete them in a later round after the regions are fully contained within a chunk. We demonstrate the algorithm by clustering an affinity graph with over 1.5 trillion edges between 135 billion supervoxels derived from a 3D electron microscopic brain image.

(2 min) In this paper, we address Novel Class Discovery (NCD), the task of unveiling new classes in a set of unlabeled samples given a labeled dataset with known classes. We exploit the peculiarities of NCD to build a new framework, named Neighborhood Contrastive Learning (NCL), to learn discriminative representations that are important to clustering performance. Our contribution is twofold. First, we find that a feature extractor trained on the labeled set generates representations in which a generic query sample and its neighbors are likely to share the same class. We exploit this observation to retrieve and aggregate pseudo-positive pairs with contrastive learning, thus encouraging the model to learn more discriminative representations. Second, we notice that most of the instances are easily discriminated by the network, contributing less to the contrastive loss. To overcome this issue, we propose to generate hard negatives by mixing labeled and unlabeled samples in the feature space. We experimentally demonstrate that these two ingredients significantly contribute to clustering performance and lead our model to outperform state-of-the-art methods by a large margin (e.g., clustering accuracy +13% on CIFAR-100 and +8% on ImageNet).

(2 min) Autonomous driving is regarded as one of the most promising remedies to shield human beings from severe crashes. To this end, 3D object detection serves as the core basis of such perception system especially for the sake of path planning, motion prediction, collision avoidance, etc. Generally, stereo or monocular images with corresponding 3D point clouds are already standard layout for 3D object detection, out of which point clouds are increasingly prevalent with accurate depth information being provided. Despite existing efforts, 3D object detection on point clouds is still in its infancy due to high sparseness and irregularity of point clouds by nature, misalignment view between camera view and LiDAR bird's eye of view for modality synergies, occlusions and scale variations at long distances, etc. Recently, profound progress has been made in 3D object detection, with a large body of literature being investigated to address this vision task. As such, we present a comprehensive review of the latest progress in this field covering all the main topics including sensors, fundamentals, and the recent state-of-the-art detection methods with their pros and cons. Furthermore, we introduce metrics and provide quantitative comparisons on popular public datasets. The avenues for future work are going to be judiciously identified after an in-deep analysis of the surveyed works. Finally, we conclude this paper.

(2 min) Trajectory prediction is one of the key capabilities for robots to safely navigate and interact with pedestrians. Critical insights from human intention and behavioral patterns need to be integrated to effectively forecast long-term pedestrian behavior. Thus, we propose a framework incorporating a Mutable Intention Filter and a Warp LSTM (MIF-WLSTM) to simultaneously estimate human intention and perform trajectory prediction. The Mutable Intention Filter is inspired by particle filtering and genetic algorithms, where particles represent intention hypotheses that can be mutated throughout the pedestrian motion. Instead of predicting sequential displacement over time, our Warp LSTM learns to generate offsets on a full trajectory predicted by a nominal intention-aware linear model, which considers the intention hypotheses during filtering process. Through experiments on a publicly available dataset, we show that our method outperforms baseline approaches and demonstrate the robust performance of our method under abnormal intention-changing scenarios. Code is available at https://github.com/tedhuang96/mifwlstm.

(2 min) In this paper, we introduce a novel semantic description approach inspired on Prototype Theory foundations. We propose a Computational Prototype Model (CPM) that encodes and stores the central semantic meaning of objects category: the semantic prototype. Also, we introduce a Prototype-based Description Model that encodes the semantic meaning of an object while describing its features using our CPM model. Our description method uses semantic prototypes computed by CNN-classifications models to create discriminative signatures that describe an object highlighting its most distinctive features within the category. Our experiments show that: i) our CPM model (semantic prototype + distance metric) is able to describe the internal semantic structure of objects categories; ii) our semantic distance metric can be understood as the object visual typicality score within a category; iii) our descriptor encoding is semantically interpretable and significantly outperforms other image global encodings in clustering and classification tasks.

(2 min) Neural shape representations have recently shown to be effective in shape analysis and reconstruction tasks. Existing neural network methods require point coordinates and corresponding normal vectors to learn the implicit level sets of the shape. Normal vectors are often not provided as raw data, therefore, approximation and reorientation are required as pre-processing stages, both of which can introduce noise. In this paper, we propose a divergence guided shape representation learning approach that does not require normal vectors as input. We show that incorporating a soft constraint on the divergence of the distance function favours smooth solutions that reliably orients gradients to match the unknown normal at each point, in some cases even better than approaches that use ground truth normal vectors directly. Additionally, we introduce a novel geometric initialization method for sinusoidal shape representation networks that further improves convergence to the desired solution. We evaluate the effectiveness of our approach on the task of surface reconstruction and show state-of-the-art performance compared to other unoriented methods and on-par performance compared to oriented methods.

(2 min) Meta-learning (ML) has emerged as a promising direction in learning models under constrained resource settings like few-shot learning. The popular approaches for ML either learn a generalizable initial model or a generic parametric optimizer through episodic training. The former approaches leverage the knowledge from a batch of tasks to learn an optimal prior. In this work, we study the importance of a batch for ML. Specifically, we first incorporate a batch episodic training regimen to improve the learning of the generic parametric optimizer. We also hypothesize that the common assumption in batch episodic training that each task in a batch has an equal contribution to learning an optimal meta-model need not be true. We propose to weight the tasks in a batch according to their "importance" in improving the meta-model's learning. To this end, we introduce a training curriculum motivated by selective focus in humans, called task attended meta-training, to weight the tasks in a batch. Task attention is a standalone module that can be integrated with any batch episodic training regimen. The comparisons of the models with their non-task-attended counterparts on complex datasets like miniImageNet and tieredImageNet validate its effectiveness.

(2 min) Existing computer vision research in categorization struggles with fine-grained attributes recognition due to the inherently high intra-class variances and low inter-class variances. SOTA methods tackle this challenge by locating the most informative image regions and rely on them to classify the complete image. The most recent work, Vision Transformer (ViT), shows its strong performance in both traditional and fine-grained classification tasks. In this work, we propose a multi-stage ViT framework for fine-grained image classification tasks, which localizes the informative image regions without requiring architectural changes using the inherent multi-head self-attention mechanism. We also introduce attention-guided augmentations for improving the model's capabilities. We demonstrate the value of our approach by experimenting with four popular fine-grained benchmarks: CUB-200-2011, Stanford Cars, Stanford Dogs, and FGVC7 Plant Pathology. We also prove our model's interpretability via qualitative results.

(2 min) Deep time series metric learning is challenging due to the difficult trade-off between temporal invariance to nonlinear distortion and discriminative power in identifying non-matching sequences. This paper proposes a novel neural network-based approach for robust yet discriminative time series classification and verification. This approach adapts a parameterized attention model to time warping for greater and more adaptive temporal invariance. It is robust against not only local but also large global distortions, so that even matching pairs that do not satisfy the monotonicity, continuity, and boundary conditions can still be successfully identified. Learning of this model is further guided by dynamic time warping to impose temporal constraints for stabilized training and higher discriminative power. It can learn to augment the inter-class variation through warping, so that similar but different classes can be effectively distinguished. We experimentally demonstrate the superiority of the proposed approach over previous non-parametric and deep models by combining it with a deep online signature verification framework, after confirming its promising behavior in single-letter handwriting classification on the Unipen dataset.

(2 min) We aim to tackle the challenging yet practical scenery image outpainting task in this work. Recently, generative adversarial learning has significantly advanced the image outpainting by producing semantic consistent content for the given image. However, the existing methods always suffer from the blurry texture and the artifacts of the generative part, making the overall outpainting results lack authenticity. To overcome the weakness, this work investigates a principle way to synthesize texture-rich results by borrowing pixels from its neighbors (\ie, reference images), named \textbf{Re}ference-\textbf{G}uided \textbf{O}utpainting (ReGO). Particularly, the ReGO designs an Adaptive Content Selection (ACS) module to transfer the pixel of reference images for texture compensating of the target one. To prevent the style of the generated part from being affected by the reference images, a style ranking loss is further proposed to augment the ReGO to synthesize style-consistent results. Extensive experiments on two popular benchmarks, NS6K~\cite{yangzx} and NS8K~\cite{wang}, well demonstrate the effectiveness of our ReGO.

(2 min) In this paper, we propose to utilize Automated Machine Learning to automatically search architecture for deepfake detection. Unlike previous works, our method benefits from the superior capability of deep learning while relieving us from the high labor cost in the manual network design process. It is experimentally proved that our proposed method not only outperforms previous non-deep learning methods but achieves comparable or even better prediction accuracy compared to previous deep learning methods. To improve the generality of our method, especially when training data and testing data are manipulated by different methods, we propose a multi-task strategy in our network learning process, making it estimate potential manipulation regions in given samples as well as predict whether the samples are real. Comparing to previous works using similar strategies, our method depends much less on prior knowledge, such as no need to know which manipulation method is utilized and whether it is utilized already. Extensive experimental results on two benchmark datasets demonstrate the effectiveness of our proposed method on deepfake detection.

(2 min) Low-power computer vision on embedded devices has many applications. This paper describes a low-power technique for the object re-identification (reID) problem: matching a query image against a gallery of previously seen images. State-of-the-art techniques rely on large, computationally-intensive Deep Neural Networks (DNNs). We propose a novel hierarchical DNN architecture that uses attribute labels in the training dataset to perform efficient object reID. At each node in the hierarchy, a small DNN identifies a different attribute of the query image. The small DNN at each leaf node is specialized to re-identify a subset of the gallery: only the images with the attributes identified along the path from the root to a leaf. Thus, a query image is re-identified accurately after processing with a few small DNNs. We compare our method with state-of-the-art object reID techniques. With a 4% loss in accuracy, our approach realizes significant resource savings: 74% less memory, 72% fewer operations, and 67% lower query latency, yielding 65% less energy consumption.

(2 min) While there exist a plethora of deep learning tools and frameworks, the fast-growing complexity of the field brings new demands and challenges, such as more flexible network design, speedy computation on distributed setting, and compatibility between different tools. In this paper, we introduce Neural Network Libraries (https://nnabla.org), a deep learning framework designed from engineer's perspective, with emphasis on usability and compatibility as its core design principles. We elaborate on each of our design principles and its merits, and validate our attempts via experiments.

(2 min) Despite the great success of GANs in images translation with different conditioned inputs such as semantic segmentation and edge maps, generating high-fidelity realistic images with reference styles remains a grand challenge in conditional image-to-image translation. This paper presents a general image translation framework that incorporates optimal transport for feature alignment between conditional inputs and style exemplars in image translation. The introduction of optimal transport mitigates the constraint of many-to-one feature matching significantly while building up accurate semantic correspondences between conditional inputs and exemplars. We design a novel unbalanced optimal transport to address the transport between features with deviational distributions which exists widely between conditional inputs and exemplars. In addition, we design a semantic-activation normalization scheme that injects style features of exemplars into the image translation process successfully. Extensive experiments over multiple image translation tasks show that our method achieves superior image translation qualitatively and quantitatively as compared with the state-of-the-art.

(2 min) Recently, image captioning has aroused great interest in both academic and industrial worlds. Most existing systems are built upon large-scale datasets consisting of image-sentence pairs, which, however, are time-consuming to construct. In addition, even for the most advanced image captioning systems, it is still difficult to realize deep image understanding. In this work, we achieve unpaired image captioning by bridging the vision and the language domains with high-level semantic information. The motivation stems from the fact that the semantic concepts with the same modality can be extracted from both images and descriptions. To further improve the quality of captions generated by the model, we propose the Semantic Relationship Explorer, which explores the relationships between semantic concepts for better understanding of the image. Extensive experiments on MSCOCO dataset show that we can generate desirable captions without paired datasets. Furthermore, the proposed approach boosts five strong baselines under the paired setting, where the most significant improvement in CIDEr score reaches 8%, demonstrating that it is effective and generalizes well to a wide range of models.

(2 min) The grade of clear cell renal cell carcinoma (ccRCC) is a critical prognostic factor, making ccRCC nuclei grading a crucial task in RCC pathology analysis. Computer-aided nuclei grading aims to improve pathologists' work efficiency while reducing their misdiagnosis rate by automatically identifying the grades of tumor nuclei within histopathological images. Such a task requires precisely segment and accurately classify the nuclei. However, most of the existing nuclei segmentation and classification methods can not handle the inter-class similarity property of nuclei grading, thus can not be directly applied to the ccRCC grading task. In this paper, we propose a Composite High-Resolution Network for ccRCC nuclei grading. Specifically, we propose a segmentation network called W-Net that can separate the clustered nuclei. Then, we recast the fine-grained classification of nuclei to two cross-category classification tasks, based on two high-resolution feature extractors (HRFEs) which are proposed for learning these two tasks. The two HRFEs share the same backbone encoder with W-Net by a composite connection so that meaningful features for the segmentation task can be inherited for the classification task. Last, a head-fusion block is applied to generate the predicted label of each nucleus. Furthermore, we introduce a dataset for ccRCC nuclei grading, containing 1000 image patches with 70945 annotated nuclei. We demonstrate that our proposed method achieves state-of-the-art performance compared to existing methods on this large ccRCC grading dataset.

(2 min) We propose a manifold matching approach to generative models which includes a distribution generator (or data generator) and a metric generator. In our framework, we view the real data set as some manifold embedded in a high-dimensional Euclidean space. The distribution generator aims at generating samples that follow some distribution condensed around the real data manifold. It is achieved by matching two sets of points using their geometric shape descriptors, such as centroid and $p$-diameter, with learned distance metric; the metric generator utilizes both real data and generated samples to learn a distance metric which is close to some intrinsic geodesic distance on the real data manifold. The produced distance metric is further used for manifold matching. The two networks are learned simultaneously during the training process. We apply the approach on both unsupervised and supervised learning tasks: in unconditional image generation task, the proposed method obtains competitive results compared with existing generative models; in super-resolution task, we incorporate the framework in perception-based models and improve visual qualities by producing samples with more natural textures. Both theoretical analysis and real data experiments guarantee the feasibility and effectiveness of the proposed framework.

(2 min) Visual information extraction (VIE) has attracted increasing attention in recent years. The existing methods usually first organized optical character recognition (OCR) results into plain texts and then utilized token-level entity annotations as supervision to train a sequence tagging model. However, it expends great annotation costs and may be exposed to label confusion, and the OCR errors will also significantly affect the final performance. In this paper, we propose a unified weakly-supervised learning framework called TCPN (Tag, Copy or Predict Network), which introduces 1) an efficient encoder to simultaneously model the semantic and layout information in 2D OCR results; 2) a weakly-supervised training strategy that utilizes only key information sequences as supervision; and 3) a flexible and switchable decoder which contains two inference modes: one (Copy or Predict Mode) is to output key information sequences of different categories by copying a token from the input or predicting one in each time step, and the other (Tag Mode) is to directly tag the input sequence in a single forward pass. Our method shows new state-of-the-art performance on several public benchmarks, which fully proves its effectiveness.

(2 min) This study is to investigate and compare the facial recognition accuracy performance of Dlib ResNet against a K-Nearest Neighbour (KNN) classifier. Particularly when used against a dataset from an Asian ethnicity as Dlib ResNet was reported to have an accuracy deficiency when it comes to Asian faces. The comparisons are both implemented on the facial vectors extracted using the Histogram of Oriented Gradients (HOG) method and use the same dataset for a fair comparison. Authentication of a user by facial recognition in an electric vehicle (EV) charging station demonstrates a practical use case for such an authentication system.

(2 min) The majority of the existing methods for non-rigid 3D surface regression from monocular 2D images require an object template or point tracks over multiple frames as an input, and are still far from real-time processing rates. In this work, we present the Isometry-Aware Monocular Generative Adversarial Network (IsMo-GAN) - an approach for direct 3D reconstruction from a single image, trained for the deformation model in an adversarial manner on a light-weight synthetic dataset. IsMo-GAN reconstructs surfaces from real images under varying illumination, camera poses, textures and shading at over 250 Hz. In multiple experiments, it consistently outperforms several approaches in the reconstruction accuracy, runtime, generalisation to unknown surfaces and robustness to occlusions. In comparison to the state-of-the-art, we reduce the reconstruction error by 10-30% including the textureless case and our surfaces evince fewer artefacts qualitatively.

(2 min) In this technical report, we introduce the methods of HIKVISION_LiDAR_Det in the challenge of waymo open dataset real-time 3D detection. Our solution for the competition are built upon Centerpoint 3D detection framework. Several variants of CenterPoint are explored, including center attention head and feature pyramid network neck. In order to achieve real time detection, methods like batchnorm merge, half-precision floating point network and GPU-accelerated voxelization process are adopted. By using these methods, our team ranks 6th among all the methods on real-time 3D detection challenge in the waymo open dataset.

(2 min) Works on lottery ticket hypothesis (LTH) and single-shot network pruning (SNIP) have raised a lot of attention currently on post-training pruning (iterative magnitude pruning), and before-training pruning (pruning at initialization). The former method suffers from an extremely large computation cost and the latter category of methods usually struggles with insufficient performance. In comparison, during-training pruning, a class of pruning methods that simultaneously enjoys the training/inference efficiency and the comparable performance, temporarily, has been less explored. To better understand during-training pruning, we quantitatively study the effect of pruning throughout training from the perspective of pruning plasticity (the ability of the pruned networks to recover the original performance). Pruning plasticity can help explain several other empirical observations about neural network pruning in literature. We further find that pruning plasticity can be substantially improved by injecting a brain-inspired mechanism called neuroregeneration, i.e., to regenerate the same number of connections as pruned. Based on the insights from pruning plasticity, we design a novel gradual magnitude pruning (GMP) method, named gradual pruning with zero-cost neuroregeneration (GraNet), and its dynamic sparse training (DST) variant (GraNet-ST). Both of them advance state of the art. Perhaps most impressively, the latter for the first time boosts the sparse-to-sparse training performance over various dense-to-sparse methods by a large margin with ResNet-50 on ImageNet. We will release all codes.

(2 min) Video Question Answering is a task which requires an AI agent to answer questions grounded in video. This task entails three key challenges: (1) understand the intention of various questions, (2) capturing various elements of the input video (e.g., object, action, causality), and (3) cross-modal grounding between language and vision information. We propose Motion-Appearance Synergistic Networks (MASN), which embed two cross-modal features grounded on motion and appearance information and selectively utilize them depending on the question's intentions. MASN consists of a motion module, an appearance module, and a motion-appearance fusion module. The motion module computes the action-oriented cross-modal joint representations, while the appearance module focuses on the appearance aspect of the input video. Finally, the motion-appearance fusion module takes each output of the motion module and the appearance module as input, and performs question-guided fusion. As a result, MASN achieves new state-of-the-art performance on the TGIF-QA and MSVD-QA datasets. We also conduct qualitative analysis by visualizing the inference results of MASN. The code is available at https://github.com/ahjeongseo/MASN-pytorch.

(2 min) Direct reconstruction methods have been developed to estimate parametric images directly from the measured PET sinograms by combining the PET imaging model and tracer kinetics in an integrated framework. Due to limited counts received, signal-to-noise-ratio (SNR) and resolution of parametric images produced by direct reconstruction frameworks are still limited. Recently supervised deep learning methods have been successfully applied to medical imaging denoising/reconstruction when large number of high-quality training labels are available. For static PET imaging, high-quality training labels can be acquired by extending the scanning time. However, this is not feasible for dynamic PET imaging, where the scanning time is already long enough. In this work, we proposed an unsupervised deep learning framework for direct parametric reconstruction from dynamic PET, which was tested on the Patlak model and the relative equilibrium Logan model. The patient's anatomical prior image, which is readily available from PET/CT or PET/MR scans, was supplied as the network input to provide a manifold constraint, and also utilized to construct a kernel layer to perform non-local feature denoising. The linear kinetic model was embedded in the network structure as a 1x1 convolution layer. The training objective function was based on the PET statistical model. Evaluations based on dynamic datasets of 18F-FDG and 11C-PiB tracers show that the proposed framework can outperform the traditional and the kernel method-based direct reconstruction methods.

(2 min) In this work we present a novel solution for Video Instance Segmentation(VIS), that is automatically generating instance level segmentation masks along with object class and tracking them in a video. Our method improves the masks from segmentation and propagation branches in an online manner using the Mask Selection Network (MSN) hence limiting the noise accumulation during mask tracking. We propose an effective design of MSN by using patch-based convolutional neural network. The network is able to distinguish between very subtle differences between the masks and choose the better masks out of the associated masks accurately. Further, we make use of temporal consistency and process the video sequences in both forward and reverse manner as a post processing step to recover lost objects. The proposed method can be used to adapt any video object segmentation method for the task of VIS. Our method achieves a score of 49.1 mAP on 2021 YouTube-VIS Challenge and was ranked third place among more than 30 global teams. Our code will be available at https://github.com/SHI-Labs/Mask-Selection-Networks.

(2 min) Convolutional Neural Networks (CNNs) have achieved remarkable success in various computer vision tasks but rely on tremendous computational cost. To solve this problem, existing approaches either compress well-trained large-scale models or learn lightweight models with carefully designed network structures. In this work, we make a close study of the convolution operator, which is the basic unit used in CNNs, to reduce its computing load. In particular, we propose a compact convolution module, called CompConv, to facilitate efficient feature learning. With the divide-and-conquer strategy, CompConv is able to save a great many computations as well as parameters to produce a certain dimensional feature map. Furthermore, CompConv discreetly integrates the input features into the outputs to efficiently inherit the input information. More importantly, the novel CompConv is a plug-and-play module that can be directly applied to modern CNN structures to replace the vanilla convolution layers without further effort. Extensive experimental results suggest that CompConv can adequately compress the benchmark CNN structures yet barely sacrifice the performance, surpassing other competitors.

(2 min) Super-resolution (SR) is a one-to-many task with multiple possible solutions. However, previous works were not concerned about this characteristic. For a one-to-many pipeline, the generator should be able to generate multiple estimates of the reconstruction, and not be penalized for generating similar and equally realistic images. To achieve this, we propose adding weighted pixel-wise noise after every Residual-in-Residual Dense Block (RRDB) to enable the generator to generate various images. We modify the strict content loss to not penalize the stochastic variation in reconstructed images as long as it has consistent content. Additionally, we observe that there are out-of-focus regions in the DIV2K, DIV8K datasets that provide unhelpful guidelines. We filter blurry regions in the training data using the method of [10]. Finally, we modify the discriminator to receive the low-resolution image as a reference image along with the target image to provide better feedback to the generator. Using our proposed methods, we were able to improve the performance of ESRGAN in x4 perceptual SR and achieve the state-of-the-art LPIPS score in x16 perceptual extreme SR.

(2 min) We propose a fully automated system that simultaneously estimates the camera intrinsics, the ground plane, and physical distances between people from a single RGB image or video captured by a camera viewing a 3-D scene from a fixed vantage point. To automate camera calibration and distance estimation, we leverage priors about human pose and develop a novel direct formulation for pose-based auto-calibration and distance estimation, which shows state-of-the-art performance on publicly available datasets. The proposed approach enables existing camera systems to measure physical distances without needing a dedicated calibration process or range sensors, and is applicable to a broad range of use cases such as social distancing and workplace safety. Furthermore, to enable evaluation and drive research in this area, we contribute to the publicly available MEVA dataset with additional distance annotations, resulting in MEVADA -- the first evaluation benchmark in the world for the pose-based auto-calibration and distance estimation problem.

(2 min) First attempts of prediction of the facial growth (FG) direction were made over half of a century ago. Despite numerous attempts and elapsed time, a satisfactory method has not been established yet and the problem still poses a challenge for medical experts. To our knowledge, this paper is the first Machine Learning approach to the prediction of FG direction. Conducted data analysis reveals the inherent complexity of the problem and explains the reasons of difficulty in FG direction prediction based on 2D X-ray images. To perform growth forecasting, we employ a wide range of algorithms, from logistic regression, through tree ensembles to neural networks and consider three, slightly different, problem formulations. The resulting classification accuracy varies between 71% and 75%.

(2 min) We study how to evaluate the quantitative information content of a region within an image for a particular label. To this end, we bridge class activation maps with information theory. We develop an informative class activation map (infoCAM). Given a classification task, infoCAM depict how to accumulate information of partial regions to that of the entire image toward a label. Thus, we can utilise infoCAM to locate the most informative features for a label. When applied to an image classification task, infoCAM performs better than the traditional classification map in the weakly supervised object localisation task. We achieve state-of-the-art results on Tiny-ImageNet.

(2 min) Hoping to stimulate new research in individual animal identification from images, we propose to formulate the problem as the human-machine Continual Curation of images and animal identities. This is an open world recognition problem, where most new animals enter the system after its algorithms are initially trained and deployed. Continual Curation, as defined here, requires (1) an improvement in the effectiveness of current recognition methods, (2) a pairwise verification algorithm that allows the possibility of no decision, and (3) an algorithmic decision mechanism that seeks human input to guide the curation process. Error metrics must evaluate the ability of recognition algorithms to identify not only animals that have been seen just once or twice but also recognize new animals not in the database. An important measure of overall system performance is accuracy as a function of the amount of human input required.

(2 min) In the interactive segmentation, users initially click on the target object to segment the main body and then provide corrections on mislabeled regions to iteratively refine the segmentation masks. Most existing methods transform these user-provided clicks into interaction maps and concatenate them with image as the input tensor. Typically, the interaction maps are determined by measuring the distance of each pixel to the clicked points, ignoring the relation between clicks and mislabeled regions. We propose a Dynamic Click Transform Network~(DCT-Net), consisting of Spatial-DCT and Feature-DCT, to better represent user interactions. Spatial-DCT transforms each user-provided click with individual diffusion distance according to the target scale, and Feature-DCT normalizes the extracted feature map to a specific distribution predicted from the clicked points. We demonstrate the effectiveness of our proposed method and achieve favorable performance compared to the state-of-the-art on three standard benchmark datasets.

(2 min) The costly process of obtaining semantic segmentation labels has driven research towards weakly supervised semantic segmentation (WSSS) methods, using only image-level, point, or box labels. The lack of dense scene representation requires methods to increase complexity to obtain additional semantic information about the scene, often done through multiple stages of training and refinement. Current state-of-the-art (SOTA) models leverage image-level labels to produce class activation maps (CAMs) which go through multiple stages of refinement before they are thresholded to make pseudo-masks for supervision. The multi-stage approach is computationally expensive, and dependency on image-level labels for CAMs generation lacks generalizability to more complex scenes. In contrary, our method offers a single-stage approach generalizable to arbitrary dataset, that is trainable from scratch, without any dependency on pre-trained backbones, classification, or separate refinement tasks. We utilize point annotations to generate reliable, on-the-fly pseudo-masks through refined and filtered features. While our method requires point annotations that are only slightly more expensive than image-level annotations, we are to demonstrate SOTA performance on benchmark datasets (PascalVOC 2012), as well as significantly outperform other SOTA WSSS methods on recent real-world datasets (CRAID, CityPersons, IAD).

(2 min) In this paper, we propose a Bayesian switching dynamical model for segmentation of 3D pose data over time that uncovers interpretable patterns in the data and is generative. Our model decomposes highly correlated skeleton data into a set of few spatial basis of switching temporal processes in a low-dimensional latent framework. We parameterize these temporal processes with regard to a switching deep vector autoregressive prior in order to accommodate both multimodal and higher-order nonlinear inter-dependencies. This results in a dynamical deep generative latent model that parses the meaningful intrinsic states in the dynamics of 3D pose data using approximate variational inference, and enables a realistic low-level dynamical generation and segmentation of complex skeleton movements. Our experiments on four biological motion data containing bat flight, salsa dance, walking, and golf datasets substantiate superior performance of our model in comparison with the state-of-the-art methods.

(2 min) In recent times, deep learning-based steganalysis classifiers became popular due to their state-of-the-art performance. Most deep steganalysis classifiers usually extract noise residuals using high-pass filters as preprocessing steps and feed them to their deep model for classification. It is observed that recent steganographic embedding does not always restrict their embedding in the high-frequency zone; instead, they distribute it as per embedding policy. Therefore, besides noise residual, learning the embedding zone is another challenging task. In this work, unlike the conventional approaches, the proposed model first extracts the noise residual using learned denoising kernels to boost the signal-to-noise ratio. After preprocessing, the sparse noise residuals are fed to a novel Multi-Contextual Convolutional Neural Network (M-CNET) that uses heterogeneous context size to learn the sparse and low-amplitude representation of noise residuals. The model performance is further improved by incorporating the Self-Attention module to focus on the areas prone to steganalytic embedding. A set of comprehensive experiments is performed to show the proposed scheme's efficacy over the prior arts. Besides, an ablation study is given to justify the contribution of various modules of the proposed architecture.

(2 min) To assist human drivers and autonomous vehicles in assessing crash risks, driving scene analysis using dash cameras on vehicles and deep learning algorithms is of paramount importance. Although these technologies are increasingly available, driving scene analysis for this purpose still remains a challenge. This is mainly due to the lack of annotated large image datasets for analyzing crash risk indicators and crash likelihood, and the lack of an effective method to extract lots of required information from complex driving scenes. To fill the gap, this paper develops a scene analysis system. The Multi-Net of the system includes two multi-task neural networks that perform scene classification to provide four labels for each scene. The DeepLab v3 and YOLO v3 are combined by the system to detect and locate risky pedestrians and the nearest vehicles. All identified information can provide the situational awareness to autonomous vehicles or human drivers for identifying crash risks from the surrounding traffic. To address the scarcity of annotated image datasets for studying traffic crashes, two completely new datasets have been developed by this paper and made available to the public, which were proved to be effective in training the proposed deep neural networks. The paper further evaluates the performance of the Multi-Net and the efficiency of the developed system. Comprehensive scene analysis is further illustrated with representative examples. Results demonstrate the effectiveness of the developed system and datasets for driving scene analysis, and their supportiveness for crash risk assessment and crash prevention.

(2 min) Autonomous Vehicles (AV) are becoming more capable of navigating in complex environments with dynamic and changing conditions. A key component that enables these intelligent vehicles to overcome such conditions and become more autonomous is the sophistication of the perception and localization systems. As part of the localization system, place recognition has benefited from recent developments in other perception tasks such as place categorization or object recognition, namely with the emergence of deep learning (DL) frameworks. This paper surveys recent approaches and methods used in place recognition, particularly those based on deep learning. The contributions of this work are twofold: surveying recent sensors such as 3D LiDARs and RADARs, applied in place recognition; and categorizing the various DL-based place recognition works into supervised, unsupervised, semi-supervised, parallel, and hierarchical categories. First, this survey introduces key place recognition concepts to contextualize the reader. Then, sensor characteristics are addressed. This survey proceeds by elaborating on the various DL-based works, presenting summaries for each framework. Some lessons learned from this survey include: the importance of NetVLAD for supervised end-to-end learning; the advantages of unsupervised approaches in place recognition, namely for cross-domain applications; or the increasing tendency of recent works to seek, not only for higher performance but also for higher efficiency.

(2 min) Detection in large-scale scenes is a challenging problem due to small objects and extreme scale variation. It is essential to focus on the image regions of small objects. In this paper, we propose a novel Adaptive Zoom (AdaZoom) network as a selective magnifier with flexible shape and focal length to adaptively zoom the focus regions for object detection. Based on policy gradient, we construct a reinforcement learning framework for focus region generation, with the reward formulated by object distributions. The scales and aspect ratios of the generated regions are adaptive to the scales and distribution of objects inside. We apply variable magnification according to the scale of the region for adaptive multi-scale detection. We further propose collaborative training to complementarily promote the performance of AdaZoom and the detection network. To validate the effectiveness, we conduct extensive experiments on VisDrone2019, UAVDT, and DOTA datasets. The experiments show AdaZoom brings a consistent and significant improvement over different detection networks, achieving state-of-the-art performance on these datasets, especially outperforming the existing methods by AP of 4.64% on Vis-Drone2019.

(2 min) Person search has recently emerged as a challenging task that jointly addresses pedestrian detection and person re-identification. Existing approaches follow a fully supervised setting where both bounding box and identity annotations are available. However, annotating identities is labor-intensive, limiting the practicability and scalability of current frameworks. This paper inventively considers weakly supervised person search with only bounding box annotations. We proposed the first framework to address this novel task, namely Context-Guided Person Search (CGPS), by investigating three levels of context clues (i.e., detection, memory and scene) in unconstrained natural images. The first two are employed to promote local and global discriminative capabilities, while the latter enhances clustering accuracy. Despite its simple design, our CGPS boosts the baseline model by 8.3% in mAP on CUHK-SYSU. Surprisingly, it even achieves comparable performance to two-step person search models, while displaying higher efficiency. Our code is available at https://github.com/ljpadam/CGPS.

(2 min) We propose to learn a generative model via entropy interpolation with a Schr\"{o}dinger Bridge. The generative learning task can be formulated as interpolating between a reference distribution and a target distribution based on the Kullback-Leibler divergence. At the population level, this entropy interpolation is characterized via an SDE on $[0,1]$ with a time-varying drift term. At the sample level, we derive our Schr\"{o}dinger Bridge algorithm by plugging the drift term estimated by a deep score estimator and a deep density ratio estimator into the Euler-Maruyama method. Under some mild smoothness assumptions of the target distribution, we prove the consistency of both the score estimator and the density ratio estimator, and then establish the consistency of the proposed Schr\"{o}dinger Bridge approach. Our theoretical results guarantee that the distribution learned by our approach converges to the target distribution. Experimental results on multimodal synthetic data and benchmark data support our theoretical findings and indicate that the generative model via Schr\"{o}dinger Bridge is comparable with state-of-the-art GANs, suggesting a new formulation of generative learning. We demonstrate its usefulness in image interpolation and image inpainting.

(3 min) Industrial processes produce a considerable volume of data and thus information. Whether it is structured sensory data or semi- to unstructured textual data, the knowledge that can be derived from it is critical to the sustainable development of the industrial process. A key challenge of this sustainability is the intelligent management of the generated data, as well as the knowledge extracted from it, in order to utilize this knowledge for improving future procedures. This challenge is a result of the tailored documentation methods and domain-specific requirements, which include the need for quick visibility of the documented knowledge. In this paper, we utilize the expert knowledge documented in chip-design failure reports in supporting user access to information that is relevant to a current chip design. Unstructured, free, textual data in previous failure documentations provides a valuable source of lessons-learned, which expert design-engineers have experienced, solved and documented. To achieve a sustainable utilization of knowledge within the company, not only the inherent knowledge has to be mined from unstructured textual data, but also the relations between the lessons-learned, uncovering potentially unknown links. In this research, a knowledge graph is constructed, in order to represent and use the interconnections between reported design failures. A search engine is developed and applied onto the graph to answer queries. In contrast to mere keyword-based searching, the searchability of the knowledge graph offers enhanced search results beyond direct matches and acts as a mean for generating explainable results and result recommendations. Results are provided to the design engineer through an interactive search interface, in which, the feedback from the user is used to further optimize relations for future iterations of the knowledge graph.

(2 min) Assessing an AI agent that can converse in human language and understand visual content is challenging. Generation metrics, such as BLEU scores favor correct syntax over semantics. Hence a discriminative approach is often used, where an agent ranks a set of candidate options. The mean reciprocal rank (MRR) metric evaluates the model performance by taking into account the rank of a single human-derived answer. This approach, however, raises a new challenge: the ambiguity and synonymy of answers, for instance, semantic equivalence (e.g., `yeah' and `yes'). To address this, the normalized discounted cumulative gain (NDCG) metric has been used to capture the relevance of all the correct answers via dense annotations. However, the NDCG metric favors the usually applicable uncertain answers such as `I don't know. Crafting a model that excels on both MRR and NDCG metrics is challenging. Ideally, an AI agent should answer a human-like reply and validate the correctness of any answer. To address this issue, we describe a two-step non-parametric ranking approach that can merge strong MRR and NDCG models. Using our approach, we manage to keep most MRR state-of-the-art performance (70.41% vs. 71.24%) and the NDCG state-of-the-art performance (72.16% vs. 75.35%). Moreover, our approach won the recent Visual Dialog 2020 challenge. Source code is available at https://github.com/idansc/mrr-ndcg.

(2 min) Over the past two decades, recommender systems have attracted a lot of interest due to the explosion in the amount of data in online applications. A particular attention has been paid to collaborative filtering, which is the most widely used in applications that involve information recommendations. Collaborative filtering (CF) uses the known preference of a group of users to make predictions and recommendations about the unknown preferences of other users (recommendations are made based on the past behavior of users). First introduced in the 1990s, a wide variety of increasingly successful models have been proposed. Due to the success of machine learning techniques in many areas, there has been a growing emphasis on the application of such algorithms in recommendation systems. In this article, we present an overview of the CF approaches for recommender systems, their two main categories, and their evaluation metrics. We focus on the application of classical Machine Learning algorithms to CF recommender systems by presenting their evolution from their first use-cases to advanced Machine Learning models. We attempt to provide a comprehensive and comparative overview of CF systems (with python implementations) that can serve as a guideline for research and practice in this area.

(2 min) Lawyers and judges spend a large amount of time researching the proper legal authority to cite while drafting decisions. In this paper, we develop a citation recommendation tool that can help improve efficiency in the process of opinion drafting. We train four types of machine learning models, including a citation-list based method (collaborative filtering) and three context-based methods (text similarity, BiLSTM and RoBERTa classifiers). Our experiments show that leveraging local textual context improves recommendation, and that deep neural models achieve decent performance. We show that non-deep text-based methods benefit from access to structured case metadata, but deep models only benefit from such access when predicting from context of insufficient length. We also find that, even after extensive training, RoBERTa does not outperform a recurrent neural model, despite its benefits of pretraining. Our behavior analysis of the RoBERTa model further shows that predictive performance is stable across time and citation classes.

(2 min) Visual information extraction (VIE) has attracted increasing attention in recent years. The existing methods usually first organized optical character recognition (OCR) results into plain texts and then utilized token-level entity annotations as supervision to train a sequence tagging model. However, it expends great annotation costs and may be exposed to label confusion, and the OCR errors will also significantly affect the final performance. In this paper, we propose a unified weakly-supervised learning framework called TCPN (Tag, Copy or Predict Network), which introduces 1) an efficient encoder to simultaneously model the semantic and layout information in 2D OCR results; 2) a weakly-supervised training strategy that utilizes only key information sequences as supervision; and 3) a flexible and switchable decoder which contains two inference modes: one (Copy or Predict Mode) is to output key information sequences of different categories by copying a token from the input or predicting one in each time step, and the other (Tag Mode) is to directly tag the input sequence in a single forward pass. Our method shows new state-of-the-art performance on several public benchmarks, which fully proves its effectiveness.

(2 min) Online gaming is growing faster than ever before, with increasing challenges of providing better user experience. Recommender systems (RS) for online games face unique challenges since they must fulfill players' distinct desires, at different user levels, based on their action sequences of various action types. Although many sequential RS already exist, they are mainly single-sequence, single-task, and single-user-level. In this paper, we introduce a new sequential recommendation model for multiple sequences, multiple tasks, and multiple user levels (abbreviated as M$^3$Rec) in Tencent Games platform, which can fully utilize complex data in online games. We leverage Graph Neural Network and multi-task learning to design M$^3$Rec in order to model the complex information in the heterogeneous sequential recommendation scenario of Tencent Games. We verify the effectiveness of M$^3$Rec on three online games of Tencent Games platform, in both offline and online evaluations. The results show that M$^3$Rec successfully addresses the challenges of recommendation in online games, and it generates superior recommendations compared with state-of-the-art sequential recommendation approaches.

(2 min) Income verification is the problem of validating a person's stated income given basic identity information such as name, location, job title and employer. It is widely used in the context of mortgage lending, rental applications and other financial risk models. However, the current processes surrounding verification involve significant human effort and document gathering which can be both time-consuming and expensive. In this paper, we propose a novel model for verifying an individual's income given very limited identity information typically available in loan applications. Our model is a combination of a deep neural network and hand-engineered features. The hand engineered features are based upon matching the input information against income records extracted automatically from various publicly available online sources (e.g. payscale.com, H-1B filings, government employee salaries). We conduct experiments on two data sets, one simulated from H-1B records and the other from a real-world data set of peer-to-peer (P2P) loan applications obtained from one of the world's largest P2P lending platform. Our results show a significant reduction in error of 3-6% relative to several strong baselines. We also perform ablation studies to demonstrate that a combined model is indeed necessary to achieve state-of-the-art performance on this task.

(2 min) Recently, Rendle has warned that the use of sampling-based top-$k$ metrics might not suffice. This throws a number of recent studies on deep learning-based recommendation algorithms, and classic non-deep-learning algorithms using such a metric, into jeopardy. In this work, we thoroughly investigate the relationship between the sampling and global top-$K$ Hit-Ratio (HR, or Recall), originally proposed by Koren[2] and extensively used by others. By formulating the problem of aligning sampling top-$k$ ($SHR@k$) and global top-$K$ ($HR@K$) Hit-Ratios through a mapping function $f$, so that $SHR@k\approx HR@f(k)$, we demonstrate both theoretically and experimentally that the sampling top-$k$ Hit-Ratio provides an accurate approximation of its global (exact) counterpart, and can consistently predict the correct winners (the same as indicate by their corresponding global Hit-Ratios).

(2 min) Obtaining an explanation for an SQL query result can enrich the analysis experience, reveal data errors, and provide deeper insight into the data. Inference query explanation seeks to explain unexpected aggregate query results on inference data; such queries are challenging to explain because an explanation may need to be derived from the source, training, or inference data in an ML pipeline. In this paper, we model an objective function as a black-box function and propose BOExplain, a novel framework for explaining inference queries using Bayesian optimization (BO). An explanation is a predicate defining the input tuples that should be removed so that the query result of interest is significantly affected. BO - a technique for finding the global optimum of a black-box function - is used to find the best predicate. We develop two new techniques (individual contribution encoding and warm start) to handle categorical variables. We perform experiments showing that the predicates found by BOExplain have a higher degree of explanation compared to those found by the state-of-the-art query explanation engines. We also show that BOExplain is effective at deriving explanations for inference queries from source and training data on a variety of real-world datasets. BOExplain is open-sourced as a Python package at https://github.com/sfu-db/BOExplain.

(2 min) Neural shape representations have recently shown to be effective in shape analysis and reconstruction tasks. Existing neural network methods require point coordinates and corresponding normal vectors to learn the implicit level sets of the shape. Normal vectors are often not provided as raw data, therefore, approximation and reorientation are required as pre-processing stages, both of which can introduce noise. In this paper, we propose a divergence guided shape representation learning approach that does not require normal vectors as input. We show that incorporating a soft constraint on the divergence of the distance function favours smooth solutions that reliably orients gradients to match the unknown normal at each point, in some cases even better than approaches that use ground truth normal vectors directly. Additionally, we introduce a novel geometric initialization method for sinusoidal shape representation networks that further improves convergence to the desired solution. We evaluate the effectiveness of our approach on the task of surface reconstruction and show state-of-the-art performance compared to other unoriented methods and on-par performance compared to oriented methods.

(2 min) Strategies for machine-learning(ML)-accelerated discovery that are general across materials composition spaces are essential, but demonstrations of ML have been primarily limited to narrow composition variations. By addressing the scarcity of data in promising regions of chemical space for challenging targets like open-shell transition-metal complexes, general representations and transferable ML models that leverage known relationships in existing data will accelerate discovery. Over a large set (ca. 1000) of isovalent transition-metal complexes, we quantify evident relationships for different properties (i.e., spin-splitting and ligand dissociation) between rows of the periodic table (i.e., 3d/4d metals and 2p/3p ligands). We demonstrate an extension to graph-based revised autocorrelation (RAC) representation (i.e., eRAC) that incorporates the effective nuclear charge alongside the nuclear charge heuristic that otherwise overestimates dissimilarity of isovalent complexes. To address the common challenge of discovery in a new space where data is limited, we introduce a transfer learning approach in which we seed models trained on a large amount of data from one row of the periodic table with a small number of data points from the additional row. We demonstrate the synergistic value of the eRACs alongside this transfer learning strategy to consistently improve model performance. Analysis of these models highlights how the approach succeeds by reordering the distances between complexes to be more consistent with the periodic table, a property we expect to be broadly useful for other materials domains.

(2 min) This paper reviews the current state of the art in Artificial Intelligence (AI) technologies and applications in the context of the creative industries. A brief background of AI, and specifically Machine Learning (ML) algorithms, is provided including Convolutional Neural Network (CNNs), Generative Adversarial Networks (GANs), Recurrent Neural Networks (RNNs) and Deep Reinforcement Learning (DRL). We categorise creative applications into five groups related to how AI technologies are used: i) content creation, ii) information analysis, iii) content enhancement and post production workflows, iv) information extraction and enhancement, and v) data compression. We critically examine the successes and limitations of this rapidly advancing technology in each of these areas. We further differentiate between the use of AI as a creative tool and its potential as a creator in its own right. We foresee that, in the near future, machine learning-based AI will be adopted widely as a tool or collaborative assistant for creativity. In contrast, we observe that the successes of machine learning in domains with fewer constraints, where AI is the `creator', remain modest. The potential of AI (or its developers) to win awards for its original creations in competition with human creatives is also limited, based on contemporary technologies. We therefore conclude that, in the context of creative industries, maximum benefit from AI will be derived where its focus is human centric -- where it is designed to augment, rather than replace, human creativity.

(2 min) We propose a flexible yet interpretable model for high-dimensional data with time-varying second order statistics, motivated and applied to functional neuroimaging data. Motivated by the neuroscience literature, we factorize the covariances into sparse spatial and smooth temporal components. While this factorization results in both parsimony and domain interpretability, the resulting estimation problem is nonconvex. To this end, we design a two-stage optimization scheme with a carefully tailored spectral initialization, combined with iteratively refined alternating projected gradient descent. We prove a linear convergence rate up to a nontrivial statistical error for the proposed descent scheme and establish sample complexity guarantees for the estimator. We further quantify the statistical error for the multivariate Gaussian case. Empirical results using simulated and real brain imaging data illustrate that our approach outperforms existing baselines.

(2 min) Numerous invariant (or equivariant) neural networks have succeeded in handling invariant data such as point clouds and graphs. However, a generalization theory for the neural networks has not been well developed, because several essential factors for the theory, such as network size and margin distribution, are not deeply connected to the invariance and equivariance. In this study, we develop a novel generalization error bound for invariant and equivariant deep neural networks. To describe the effect of invariance and equivariance on generalization, we develop a notion of a \textit{quotient feature space}, which measures the effect of group actions for the properties. Our main result proves that the volume of quotient feature spaces can describe the generalization error. Furthermore, the bound shows that the invariance and equivariance significantly improve the leading term of the bound. We apply our result to specific invariant and equivariant networks, such as DeepSets (Zaheer et al. (2017)), and show that their generalization bound is considerably improved by $\sqrt{n!}$, where $n!$ is the number of permutations. We also discuss the expressive power of invariant DNNs and show that they can achieve an optimal approximation rate. Our experimental result supports our theoretical claims.

(2 min) Studying the relationship between neuroanatomy and cognitive decline due to Alzheimer's has been a major research focus in the last decade. However, to infer cause-effect relationships rather than simple associations from observational data, we need to (i) express the causal relationships leading to cognitive decline in a graphical model, and (ii) ensure the causal effect of interest is identifiable from the collected data. We derive a causal graph from the current clinical knowledge on cause and effect in the Alzheimer's disease continuum, and show that identifiability of the causal effect requires all confounders to be known and measured. However, in complex neuroimaging studies, we neither know all potential confounders nor do we have data on them. To alleviate this requirement, we leverage the dependencies among multiple causes by deriving a substitute confounder via a probabilistic latent factor model. In our theoretical analysis, we prove that using the substitute confounder enables identifiability of the causal effect of neuroanatomy on cognition. We quantitatively evaluate the effectiveness of our approach on semi-synthetic data, where we know the true causal effects, and illustrate its use on real data on the Alzheimer's disease continuum, where it reveals important causes that otherwise would have been missed.

(2 min) Self- and mutually-exciting point processes are popular models in machine learning and statistics for dependent discrete event data. To date, most existing models assume stationary kernels (including the classical Hawkes processes) and simple parametric models. Modern applications with complex event data require more general point process models that can incorporate contextual information of the events, called marks, besides the temporal and location information. Moreover, such applications often require non-stationary models to capture more complex spatio-temporal dependence. To tackle these challenges, a key question is to devise a versatile influence kernel in the point process model. In this paper, we introduce a novel and general neural network-based non-stationary influence kernel with high expressiveness for handling complex discrete events data while providing theoretical performance guarantees. We demonstrate the superior performance of our proposed method compared with the state-of-the-art on synthetic and real data.

(2 min) In an ever expanding set of research and application areas, deep neural networks (DNNs) set the bar for algorithm performance. However, depending upon additional constraints such as processing power and execution time limits, or requirements such as verifiable safety guarantees, it may not be feasible to actually use such high-performing DNNs in practice. Many techniques have been developed in recent years to compress or distill complex DNNs into smaller, faster or more understandable models and controllers. This work seeks to identify reduced models that not only preserve a desired performance level, but also, for example, succinctly explain the latent knowledge represented by a DNN. We illustrate the effectiveness of the proposed approach on the evaluation of decision tree variants and kernel machines in the context of benchmark reinforcement learning tasks.

(2 min) Most explanation methods in deep learning map importance estimates for a model's prediction back to the original input space. These "visual" explanations are often insufficient, as the model's actual concept remains elusive. Moreover, without insights into the model's semantic concept, it is difficult -- if not impossible -- to intervene on the model's behavior via its explanations, called Explanatory Interactive Learning. Consequently, we propose to intervene on a Neuro-Symbolic scene representation, which allows one to revise the model on the semantic level, e.g. "never focus on the color to make your decision". We compiled a novel confounded visual scene data set, the CLEVR-Hans data set, capturing complex compositions of different objects. The results of our experiments on CLEVR-Hans demonstrate that our semantic explanations, i.e. compositional explanations at a per-object level, can identify confounders that are not identifiable using "visual" explanations only. More importantly, feedback on this semantic level makes it possible to revise the model from focusing on these factors.

(2 min) Human personality traits are the key drivers behind our decision-making, influencing our life path on a daily basis. Inference of personality traits, such as Myers-Briggs Personality Type, as well as an understanding of dependencies between personality traits and users' behavior on various social media platforms is of crucial importance to modern research and industry applications. The emergence of diverse and cross-purpose social media avenues makes it possible to perform user personality profiling automatically and efficiently based on data represented across multiple data modalities. However, the research efforts on personality profiling from multi-source multi-modal social media data are relatively sparse, and the level of impact of different social network data on machine learning performance has yet to be comprehensively evaluated. Furthermore, there is not such dataset in the research community to benchmark. This study is one of the first attempts towards bridging such an important research gap. Specifically, in this work, we infer the Myers-Briggs Personality Type indicators, by applying a novel multi-view fusion framework, called "PERS" and comparing the performance results not just across data modalities but also with respect to different social network data sources. Our experimental results demonstrate the PERS's ability to learn from multi-view data for personality profiling by efficiently leveraging on the significantly different data arriving from diverse social multimedia sources. We have also found that the selection of a machine learning approach is of crucial importance when choosing social network data sources and that people tend to reveal multiple facets of their personality in different social media avenues. Our released social multimedia dataset facilitates future research on this direction.

(2 min) Given the increasing importance of machine learning (ML) in our lives, several algorithmic fairness techniques have been proposed to mitigate biases in the outcomes of the ML models. However, most of these techniques are specialized to cater to a single family of ML models and a specific definition of fairness, limiting their adaptibility in practice. We introduce a general constrained Bayesian optimization (BO) framework to optimize the performance of any ML model while enforcing one or multiple fairness constraints. BO is a model-agnostic optimization method that has been successfully applied to automatically tune the hyperparameters of ML models. We apply BO with fairness constraints to a range of popular models, including random forests, gradient boosting, and neural networks, showing that we can obtain accurate and fair solutions by acting solely on the hyperparameters. We also show empirically that our approach is competitive with specialized techniques that enforce model-specific fairness constraints, and outperforms preprocessing methods that learn fair representations of the input data. Moreover, our method can be used in synergy with such specialized fairness techniques to tune their hyperparameters. Finally, we study the relationship between fairness and the hyperparameters selected by BO. We observe a correlation between regularization and unbiased models, explaining why acting on the hyperparameters leads to ML models that generalize well and are fair.

(2 min) Graph neural networks (GNNs) have attracted increasing interests. With broad deployments of GNNs in real-world applications, there is an urgent need for understanding the robustness of GNNs under adversarial attacks, especially in realistic setups. In this work, we study the problem of attacking GNNs in a restricted and realistic setup, by perturbing the features of a small set of nodes, with no access to model parameters and model predictions. Our formal analysis draws a connection between this type of attacks and an influence maximization problem on the graph. This connection not only enhances our understanding on the problem of adversarial attack on GNNs, but also allows us to propose a group of effective and practical attack strategies. Our experiments verify that the proposed attack strategies significantly degrade the performance of three popular GNN models and outperform baseline adversarial attack strategies.

(3 min) The COVID-19 pandemic has impacted lives and economies across the globe, leading to many deaths. While vaccination is an important intervention, its roll-out is slow and unequal across the globe. Therefore, extensive testing still remains one of the key methods to monitor and contain the virus. Testing on a large scale is expensive and arduous. Hence, we need alternate methods to estimate the number of cases. Online surveys have been shown to be an effective method for data collection amidst the pandemic. In this work, we develop machine learning models to estimate the prevalence of COVID-19 using self-reported symptoms. Our best model predicts the daily cases with a mean absolute error (MAE) of 226.30 (normalized MAE of 27.09%) per state, which demonstrates the possibility of predicting the actual number of confirmed cases by utilizing self-reported symptoms. The models are developed at two levels of data granularity - local models, which are trained at the state level, and a single global model which is trained on the combined data aggregated across all states. Our results indicate a lower error on the local models as opposed to the global model. In addition, we also show that the most important symptoms (features) vary considerably from state to state. This work demonstrates that the models developed on crowd-sourced data, curated via online platforms, can complement the existing epidemiological surveillance infrastructure in a cost-effective manner. The code is publicly available at https://github.com/parthpatwa/Can-Self-Reported-Symptoms-Predict-Daily-COVID-19-Cases.

(2 min) We propose firefly neural architecture descent, a general framework for progressively and dynamically growing neural networks to jointly optimize the networks' parameters and architectures. Our method works in a steepest descent fashion, which iteratively finds the best network within a functional neighborhood of the original network that includes a diverse set of candidate network structures. By using Taylor approximation, the optimal network structure in the neighborhood can be found with a greedy selection procedure. We show that firefly descent can flexibly grow networks both wider and deeper, and can be applied to learn accurate but resource-efficient neural architectures that avoid catastrophic forgetting in continual learning. Empirically, firefly descent achieves promising results on both neural architecture search and continual learning. In particular, on a challenging continual image classification task, it learns networks that are smaller in size but have higher average accuracy than those learned by the state-of-the-art methods.

(2 min) As the analytic tools become more powerful, and more data are generated on a daily basis, the issue of data privacy arises. This leads to the study of the design of privacy-preserving machine learning algorithms. Given two objectives, namely, utility maximization and privacy-loss minimization, this work is based on two previously non-intersecting regimes -- Compressive Privacy and multi-kernel method. Compressive Privacy is a privacy framework that employs utility-preserving lossy-encoding scheme to protect the privacy of the data, while multi-kernel method is a kernel based machine learning regime that explores the idea of using multiple kernels for building better predictors. The compressive multi-kernel method proposed consists of two stages -- the compression stage and the multi-kernel stage. The compression stage follows the Compressive Privacy paradigm to provide the desired privacy protection. Each kernel matrix is compressed with a lossy projection matrix derived from the Discriminant Component Analysis (DCA). The multi-kernel stage uses the signal-to-noise ratio (SNR) score of each kernel to non-uniformly combine multiple compressive kernels. The proposed method is evaluated on two mobile-sensing datasets -- MHEALTH and HAR -- where activity recognition is defined as utility and person identification is defined as privacy. The results show that the compression regime is successful in privacy preservation as the privacy classification accuracies are almost at the random-guess level in all experiments. On the other hand, the novel SNR-based multi-kernel shows utility classification accuracy improvement upon the state-of-the-art in both datasets. These results indicate a promising direction for research in privacy-preserving machine learning.

(2 min) Assessing an AI agent that can converse in human language and understand visual content is challenging. Generation metrics, such as BLEU scores favor correct syntax over semantics. Hence a discriminative approach is often used, where an agent ranks a set of candidate options. The mean reciprocal rank (MRR) metric evaluates the model performance by taking into account the rank of a single human-derived answer. This approach, however, raises a new challenge: the ambiguity and synonymy of answers, for instance, semantic equivalence (e.g., `yeah' and `yes'). To address this, the normalized discounted cumulative gain (NDCG) metric has been used to capture the relevance of all the correct answers via dense annotations. However, the NDCG metric favors the usually applicable uncertain answers such as `I don't know. Crafting a model that excels on both MRR and NDCG metrics is challenging. Ideally, an AI agent should answer a human-like reply and validate the correctness of any answer. To address this issue, we describe a two-step non-parametric ranking approach that can merge strong MRR and NDCG models. Using our approach, we manage to keep most MRR state-of-the-art performance (70.41% vs. 71.24%) and the NDCG state-of-the-art performance (72.16% vs. 75.35%). Moreover, our approach won the recent Visual Dialog 2020 challenge. Source code is available at https://github.com/idansc/mrr-ndcg.

(2 min) Behavioral changes are the earliest signs of a mental disorder, but arguably, the dynamics of brain function gets affected even earlier. Subsequently, spatio-temporal structure of disorder-specific dynamics is crucial for early diagnosis and understanding the disorder mechanism. A common way of learning discriminatory features relies on training a classifier and evaluating feature importance. Classical classifiers, based on handcrafted features are quite powerful, but suffer the curse of dimensionality when applied to large input dimensions of spatio-temporal data. Deep learning algorithms could handle the problem and a model introspection could highlight discriminatory spatio-temporal regions but need way more samples to train. In this paper we present a novel self supervised training schema which reinforces whole sequence mutual information local to context (whole MILC). We pre-train the whole MILC model on unlabeled and unrelated healthy control data. We test our model on three different disorders (i) Schizophrenia (ii) Autism and (iii) Alzheimers and four different studies. Our algorithm outperforms existing self-supervised pre-training methods and provides competitive classification results to classical machine learning algorithms. Importantly, whole MILC enables attribution of subject diagnosis to specific spatio-temporal regions in the fMRI signal.

(2 min) Glaucoma is the leading cause of irreversible blindness in the world, affecting over 70 million people. The cumbersome Standard Automated Perimetry (SAP) test is most frequently used to detect visual loss due to glaucoma. Due to the SAP test's innate difficulty and its high test-retest variability, we propose the RetiNerveNet, a deep convolutional recursive neural network for obtaining estimates of the SAP visual field. RetiNerveNet uses information from the more objective Spectral-Domain Optical Coherence Tomography (SDOCT). RetiNerveNet attempts to trace-back the arcuate convergence of the retinal nerve fibers, starting from the Retinal Nerve Fiber Layer (RNFL) thickness around the optic disc, to estimate individual age-corrected 24-2 SAP values. Recursive passes through the proposed network sequentially yield estimates of the visual locations progressively farther from the optic disc. While all the methods used for our experiments exhibit lower performance for the advanced disease group, the proposed network is observed to be more accurate than all the baselines for estimating the individual visual field values. We further augment RetiNerveNet to additionally predict the SAP Mean Deviation values and also create an ensemble of RetiNerveNets that further improves the performance, by increasingly weighting-up underrepresented parts of the training data.

(2 min) We present a principled approach to incorporating labels in VAEs that captures the rich characteristic information associated with those labels. While prior work has typically conflated these by learning latent variables that directly correspond to label values, we argue this is contrary to the intended effect of supervision in VAEs-capturing rich label characteristics with the latents. For example, we may want to capture the characteristics of a face that make it look young, rather than just the age of the person. To this end, we develop the CCVAE, a novel VAE model and concomitant variational objective which captures label characteristics explicitly in the latent space, eschewing direct correspondences between label values and latents. Through judicious structuring of mappings between such characteristic latents and labels, we show that the CCVAE can effectively learn meaningful representations of the characteristics of interest across a variety of supervision schemes. In particular, we show that the CCVAE allows for more effective and more general interventions to be performed, such as smooth traversals within the characteristics for a given label, diverse conditional generation, and transferring characteristics across datapoints.

(2 min) We study the algorithmic stability of Nesterov's accelerated gradient method. For convex quadratic objectives, Chen et al. (2018) proved that the uniform stability of the method grows quadratically with the number of optimization steps, and conjectured that the same is true for the general convex and smooth case. We disprove this conjecture and show, for two notions of algorithmic stability (including uniform stability), that the stability of Nesterov's accelerated method in fact deteriorates exponentially fast with the number of gradient steps. This stands in sharp contrast to the bounds in the quadratic case, but also to known results for non-accelerated gradient methods where stability typically grows linearly with the number of steps.

(2 min) When dealing with deep neural network (DNN) applications on edge devices, continuously updating the model is important. Although updating a model with real incoming data is ideal, using all of them is not always feasible due to limits, such as labeling and communication costs. Thus, it is necessary to filter and select the data to use for training (i.e., active learning) on the device. In this paper, we formalize a practical active learning problem for DNNs on edge devices and propose a general task-agnostic framework to tackle this problem, which reduces it to a stream submodular maximization. This framework is light enough to be run with low computational resources, yet provides solutions whose quality is theoretically guaranteed thanks to the submodular property. Through this framework, we can configure data selection criteria flexibly, including using methods proposed in previous active learning studies. We evaluate our approach on both classification and object detection tasks in a practical setting to simulate a real-life scenario. The results of our study show that the proposed framework outperforms all other methods in both tasks, while running at a practical speed on real devices.

(2 min) While there exist a plethora of deep learning tools and frameworks, the fast-growing complexity of the field brings new demands and challenges, such as more flexible network design, speedy computation on distributed setting, and compatibility between different tools. In this paper, we introduce Neural Network Libraries (https://nnabla.org), a deep learning framework designed from engineer's perspective, with emphasis on usability and compatibility as its core design principles. We elaborate on each of our design principles and its merits, and validate our attempts via experiments.

(2 min) The rise of algorithmic decision-making has spawned much research on fair machine learning (ML). Financial institutions use ML for building risk scorecards that support a range of credit-related decisions. Yet, the literature on fair ML in credit scoring is scarce. The paper makes three contributions. First, we revisit statistical fairness criteria and examine their adequacy for credit scoring. Second, we catalog algorithmic options for incorporating fairness goals in the ML model development pipeline. Last, we empirically compare different fairness processors in a profit-oriented credit scoring context using real-world data. The empirical results substantiate the evaluation of fairness measures, identify suitable options to implement fair credit scoring, and clarify the profit-fairness trade-off in lending decisions. We find that multiple fairness criteria can be approximately satisfied at once and recommend separation as a proper criterion for measuring the fairness of a scorecard. We also find fair in-processors to deliver a good balance between profit and fairness and show that algorithmic discrimination can be reduced to a reasonable level at a relatively low cost. The codes corresponding to the paper are available on GitHub.

(2 min) Real-time adaptation is imperative to the control of robots operating in complex, dynamic environments. Adaptive control laws can endow even nonlinear systems with good trajectory tracking performance, provided that any uncertain dynamics terms are linearly parameterizable with known nonlinear features. However, it is often difficult to specify such features a priori, such as for aerodynamic disturbances on rotorcraft or interaction forces between a manipulator arm and various objects. In this paper, we turn to data-driven modeling with neural networks to learn, offline from past data, an adaptive controller with an internal parametric model of these nonlinear features. Our key insight is that we can better prepare the controller for deployment with control-oriented meta-learning of features in closed-loop simulation, rather than regression-oriented meta-learning of features to fit input-output data. Specifically, we meta-learn the adaptive controller with closed-loop tracking simulation as the base-learner and the average tracking error as the meta-objective. With a nonlinear planar rotorcraft subject to wind, we demonstrate that our adaptive controller outperforms other controllers trained with regression-oriented meta-learning when deployed in closed-loop for trajectory tracking control.

(2 min) The phenomenon of adversarial examples illustrates one of the most basic vulnerabilities of deep neural networks. Among the variety of techniques introduced to surmount this inherent weakness, adversarial training has emerged as the most common and efficient strategy to achieve robustness. Typically, this is achieved by balancing robust and natural objectives. In this work, we aim to achieve better trade-off between robust and natural performances by enforcing a domain-invariant feature representation. We present a new adversarial training method, Domain Invariant Adversarial Learning (DIAL), which learns a feature representation which is both robust and domain invariant. DIAL uses a variant of Domain Adversarial Neural Network (DANN) on the natural domain and its corresponding adversarial domain. In a case where the source domain consists of natural examples and the target domain is the adversarially perturbed examples, our method learns a feature representation constrained not to discriminate between the natural and adversarial examples, and can therefore achieve a more robust representation. Our experiments indicate that our method improves both robustness and natural accuracy, when compared to current state-of-the-art adversarial training methods.

(2 min) It is widely accepted that reasoning about object shape is important for object recognition. However, the most powerful object recognition methods today do not explicitly make use of object shape during learning. In this work, motivated by recent developments in low-shot learning, findings in developmental psychology, and the increased use of synthetic data in computer vision research, we investigate how reasoning about 3D shape can be used to improve low-shot learning methods' generalization performance. We propose a new way to improve existing low-shot learning approaches by learning a discriminative embedding space using 3D object shape, and using this embedding by learning how to map images into it. Our new approach improves the performance of image-only low-shot learning approaches on multiple datasets. We also introduce Toys4K, a 3D object dataset with the largest number of object categories currently available, which supports low-shot learning.

(2 min) Topological Data Analysis (TDA) has emerged recently as a robust tool to extract and compare the structure of datasets. TDA identifies features in data such as connected components and holes and assigns a quantitative measure to these features. Several studies reported that topological features extracted by TDA tools provide unique information about the data, discover new insights, and determine which feature is more related to the outcome. On the other hand, the overwhelming success of deep neural networks in learning patterns and relationships has been proven on a vast array of data applications, images in particular. To capture the characteristics of both powerful tools, we propose \textit{TDA-Net}, a novel ensemble network that fuses topological and deep features for the purpose of enhancing model generalizability and accuracy. We apply the proposed \textit{TDA-Net} to a critical application, which is the automated detection of COVID-19 from CXR images. The experimental results showed that the proposed network achieved excellent performance and suggests the applicability of our method in practice.

(2 min) We study the problem of identifying the best arm in a stochastic multi-armed bandit game. Given a set of $n$ arms indexed from $1$ to $n$, each arm $i$ is associated with an unknown reward distribution supported on $[0,1]$ with mean $\theta_i$ and variance $\sigma_i^2$. Assume $\theta_1 > \theta_2 \geq \cdots \geq\theta_n$. We propose an adaptive algorithm which explores the gaps and variances of the rewards of the arms and makes future decisions based on the gathered information using a novel approach called \textit{grouped median elimination}. The proposed algorithm guarantees to output the best arm with probability $(1-\delta)$ and uses at most $O \left(\sum_{i = 1}^n \left(\frac{\sigma_i^2}{\Delta_i^2} + \frac{1}{\Delta_i}\right)(\ln \delta^{-1} + \ln \ln \Delta_i^{-1})\right)$ samples, where $\Delta_i$ ($i \geq 2$) denotes the reward gap between arm $i$ and the best arm and we define $\Delta_1 = \Delta_2$. This achieves a significant advantage over the variance-independent algorithms in some favorable scenarios and is the first result that removes the extra $\ln n$ factor on the best arm compared with the state-of-the-art. We further show that $\Omega \left( \sum_{i = 1}^n \left( \frac{\sigma_i^2}{\Delta_i^2} + \frac{1}{\Delta_i} \right) \ln \delta^{-1} \right)$ samples are necessary for an algorithm to achieve the same goal, thereby illustrating that our algorithm is optimal up to doubly logarithmic terms.

(2 min) By computing the rank correlation between attention weights and feature-additive explanation methods, previous analyses either invalidate or support the role of attention-based explanations as a faithful and plausible measure of salience. To investigate whether this approach is appropriate, we compare LIME, Integrated Gradients, DeepLIFT, Grad-SHAP, Deep-SHAP, and attention-based explanations, applied to two neural architectures trained on single- and pair-sequence language tasks. In most cases, we find that none of our chosen methods agree. Based on our empirical observations and theoretical objections, we conclude that rank correlation does not measure the quality of feature-additive methods. Practitioners should instead use the numerous and rigorous diagnostic methods proposed by the community.

(2 min) Deep learning techniques have achieved remarkable performance in wide-ranging tasks. However, when trained on privacy-sensitive datasets, the model parameters may expose private information in training data. Prior attempts for differentially private training, although offering rigorous privacy guarantees, lead to much lower model performance than the non-private ones. Besides, different runs of the same training algorithm produce models with large performance variance. To address these issues, we propose DPlis--Differentially Private Learning wIth Smoothing. The core idea of DPlis is to construct a smooth loss function that favors noise-resilient models lying in large flat regions of the loss landscape. We provide theoretical justification for the utility improvements of DPlis. Extensive experiments also demonstrate that DPlis can effectively boost model quality and training stability under a given privacy budget.

(2 min) We utilize classical facts from topology to show that the classification problem in machine learning is always solvable under very mild conditions. Furthermore, we show that a softmax classification network acts on an input topological space by a finite sequence of topological moves to achieve the classification task. Moreover, given a training dataset, we show how topological formalism can be used to suggest the appropriate architectural choices for neural networks designed to be trained as classifiers on the data. Finally, we show how the architecture of a neural network cannot be chosen independently from the shape of the underlying data. To demonstrate these results, we provide example datasets and show how they are acted upon by neural nets from this topological perspective.

(2 min) Although the quest for more accurate solutions is pushing deep learning research towards larger and more complex algorithms, edge devices demand efficient inference and therefore reduction in model size, latency and energy consumption. One technique to limit model size is quantization, which implies using fewer bits to represent weights and biases. Such an approach usually results in a decline in performance. Here, we introduce a method for designing optimally heterogeneously quantized versions of deep neural network models for minimum-energy, high-accuracy, nanosecond inference and fully automated deployment on chip. With a per-layer, per-parameter type automatic quantization procedure, sampling from a wide range of quantizers, model energy consumption and size are minimized while high accuracy is maintained. This is crucial for the event selection procedure in proton-proton collisions at the CERN Large Hadron Collider, where resources are strictly limited and a latency of ${\mathcal O}(1)~\mu$s is required. Nanosecond inference and a resource consumption reduced by a factor of 50 when implemented on field-programmable gate array hardware are achieved.

(2 min) One of the important and tedious task in agricultural practices is the detection of the disease on crops. It requires huge time as well as skilled labor. This paper proposes a smart and efficient technique for detection of crop disease which uses computer vision and machine learning techniques. The proposed system is able to detect 20 different diseases of 5 common plants with 93% accuracy.

(2 min) In recent years, algorithms and neural architectures based on the Weisfeiler-Leman algorithm, a well-known heuristic for the graph isomorphism problem, emerged as a powerful tool for (supervised) machine learning with graphs and relational data. Here, we give a comprehensive overview of the algorithm's use in a machine learning setting. We discuss the theoretical background, show how to use it for supervised graph- and node classification, discuss recent extensions, and its connection to neural architectures. Moreover, we give an overview of current applications and future directions to stimulate research.

(2 min) Deep neural networks that yield human interpretable decisions by architectural design have lately become an increasingly popular alternative to post hoc interpretation of traditional black-box models. Among these networks, the arguably most widespread approach is so-called prototype learning, where similarities to learned latent prototypes serve as the basis of classifying an unseen data point. In this work, we point to an important shortcoming of such approaches. Namely, there is a semantic gap between similarity in latent space and similarity in input space, which can corrupt interpretability. We design two experiments that exemplify this issue on the so-called ProtoPNet. Specifically, we find that this network's interpretability mechanism can be led astray by intentionally crafted or even JPEG compression artefacts, which can produce incomprehensible decisions. We argue that practitioners ought to have this shortcoming in mind when deploying prototype-based models in practice.

(2 min) Various iterative reconstruction algorithms for inverse problems can be unfolded as neural networks. Empirically, this approach has often led to improved results, but theoretical guarantees are still scarce. While some progress on generalization properties of neural networks have been made, great challenges remain. In this chapter, we discuss and combine these topics to present a generalization error analysis for a class of neural networks suitable for sparse reconstruction from few linear measurements. The hypothesis class considered is inspired by the classical iterative soft-thresholding algorithm (ISTA). The neural networks in this class are obtained by unfolding iterations of ISTA and learning some of the weights. Based on training samples, we aim at learning the optimal network parameters via empirical risk minimization and thereby the optimal network that reconstructs signals from their compressive linear measurements. In particular, we may learn a sparsity basis that is shared by all of the iterations/layers and thereby obtain a new approach for dictionary learning. For this class of networks, we present a generalization bound, which is based on bounding the Rademacher complexity of hypothesis classes consisting of such deep networks via Dudley's integral. Remarkably, under realistic conditions, the generalization error scales only logarithmically in the number of layers, and at most linear in number of measurements.

(2 min) Recent years saw a plethora of work on explaining complex intelligent agents. One example is the development of several algorithms that generate saliency maps which show how much each pixel attributed to the agents' decision. However, most evaluations of such saliency maps focus on image classification tasks. As far as we know, there is no work that thoroughly compares different saliency maps for Deep Reinforcement Learning agents. This paper compares four perturbation-based approaches to create saliency maps for Deep Reinforcement Learning agents trained on four different Atari 2600 games. All four approaches work by perturbing parts of the input and measuring how much this affects the agent's output. The approaches are compared using three computational metrics: dependence on the learned parameters of the agent (sanity checks), faithfulness to the agent's reasoning (input degradation), and run-time. In particular, during the sanity checks we find issues with two approaches and propose a solution to fix one of those issues.

(2 min) In this paper, we present a novel approach to the audio-visual video parsing (AVVP) task that demarcates events from a video separately for audio and visual modalities. The proposed parsing approach simultaneously detects the temporal boundaries in terms of start and end times of such events. We show how AVVP can benefit from the following techniques geared towards effective cross-modal learning: (i) adversarial training and skip connections (ii) global context aware attention and, (iii) self-supervised pretraining using an audio-video grounding objective to obtain cross-modal audio-video representations. We present extensive experimental evaluations on the Look, Listen, and Parse (LLP) dataset and show that we outperform the state-of-the-art Hybrid Attention Network (HAN) on all five metrics proposed for AVVP. We also present several ablations to validate the effect of pretraining, global attention and adversarial training.

(2 min) In the era of big data, methods for improving memory and computational efficiency have become crucial for successful deployment of technologies. Hashing is one of the most effective approaches to deal with computational limitations that come with big data. One natural way for formulating this problem is spectral hashing that directly incorporates affinity to learn binary codes. However, due to binary constraints, the optimization becomes intractable. To mitigate this challenge, different relaxation approaches have been proposed to reduce the computational load of obtaining binary codes and still attain a good solution. The problem with all existing relaxation methods is resorting to one or more additional auxiliary variables to attain high quality binary codes while relaxing the problem. The existence of auxiliary variables leads to coordinate descent approach which increases the computational complexity. We argue that introducing these variables is unnecessary. To this end, we propose a novel relaxed formulation for spectral hashing that adds no additional variables to the problem. Furthermore, instead of solving the problem in original space where number of variables is equal to the data points, we solve the problem in a much smaller space and retrieve the binary codes from this solution. This trick reduces both the memory and computational complexity at the same time. We apply two optimization techniques, namely projected gradient and optimization on manifold, to obtain the solution. Using comprehensive experiments on four public datasets, we show that the proposed efficient spectral hashing (ESH) algorithm achieves highly competitive retrieval performance compared with state of the art at low complexity.

(2 min) Security-sensitive applications that rely on Deep Neural Networks (DNNs) are vulnerable to small perturbations that are crafted to generate Adversarial Examples(AEs). The AEs are imperceptible to humans and cause DNN to misclassify them. Many defense and detection techniques have been proposed. Model's confidences and Dropout, as a popular way to estimate the model's uncertainty, have been used for AE detection but they showed limited success against black- and gray-box attacks. Moreover, the state-of-the-art detection techniques have been designed for specific attacks or broken by others, need knowledge about the attacks, are not consistent, increase model parameters overhead, are time-consuming, or have latency in inference time. To trade off these factors, we revisit the model's uncertainty and confidences and propose a novel unsupervised ensemble AE detection mechanism that 1) uses the uncertainty method called SelectiveNet, 2) processes model layers outputs, i.e.feature maps, to generate new confidence probabilities. The detection method is called Selective and Feature based Adversarial Detection (SFAD). Experimental results show that the proposed approach achieves better performance against black- and gray-box attacks than the state-of-the-art methods and achieves comparable performance against white-box attacks. Moreover, results show that SFAD is fully robust against High Confidence Attacks (HCAs) for MNIST and partially robust for CIFAR10 datasets.

(2 min) In this paper we introduce a numerical method for nonlinear parabolic PDEs that combines operator splitting with deep learning. It divides the PDE approximation problem into a sequence of separate learning problems. Since the computational graph for each of the subproblems is comparatively small, the approach can handle extremely high-dimensional PDEs. We test the method on different examples from physics, stochastic control and mathematical finance. In all cases, it yields very good results in up to 10,000 dimensions with short run times.

(2 min) Many real-world systems can be expressed in temporal networks with nodes playing far different roles in structure and function and edges representing the relationships between nodes. Identifying critical nodes can help us control the spread of public opinions or epidemics, predict leading figures in academia, conduct advertisements for various commodities, and so on. However, it is rather difficult to identify critical nodes because the network structure changes over time in temporal networks. In this paper, considering the sequence topological information of temporal networks, a novel and effective learning framework based on the combination of special GCNs and RNNs is proposed to identify nodes with the best spreading ability. The effectiveness of the approach is evaluated by weighted Susceptible-Infected-Recovered model. Experimental results on four real-world temporal networks demonstrate that the proposed method outperforms both traditional and deep learning benchmark methods in terms of the Kendall $\tau$ coefficient and top $k$ hit rate.

(2 min) Can a computer determine a piano player's skill level? Is it preferable to base this assessment on visual analysis of the player's performance or should we trust our ears over our eyes? Since current CNNs have difficulty processing long video videos, how can shorter clips be sampled to best reflect the players skill level? In this work, we collect and release a first-of-its-kind dataset for multimodal skill assessment focusing on assessing piano player's skill level, answer the asked questions, initiate work in automated evaluation of piano playing skills and provide baselines for future work. Dataset is available from: https://github.com/ParitoshParmar/Piano-Skills-Assessment.

(2 min) Image captioning has recently demonstrated impressive progress largely owing to the introduction of neural network algorithms trained on curated dataset like MS-COCO. Often work in this field is motivated by the promise of deployment of captioning systems in practical applications. However, the scarcity of data and contexts in many competition datasets renders the utility of systems trained on these datasets limited as an assistive technology in real-world settings, such as helping visually impaired people navigate and accomplish everyday tasks. This gap motivated the introduction of the novel VizWiz dataset, which consists of images taken by the visually impaired and captions that have useful, task-oriented information. In an attempt to help the machine learning computer vision field realize its promise of producing technologies that have positive social impact, the curators of the VizWiz dataset host several competitions, including one for image captioning. This work details the theory and engineering from our winning submission to the 2020 captioning competition. Our work provides a step towards improved assistive image captioning systems.

(2 min) Reinforcement learning (RL) enables robots to learn skills from interactions with the real world. In practice, the unstructured step-based exploration used in Deep RL -- often very successful in simulation -- leads to jerky motion patterns on real robots. Consequences of the resulting shaky behavior are poor exploration, or even damage to the robot. We address these issues by adapting state-dependent exploration (SDE) to current Deep RL algorithms. To enable this adaptation, we propose two extensions to the original SDE, using more general features and re-sampling the noise periodically, which leads to a new exploration method generalized state-dependent exploration (gSDE). We evaluate gSDE both in simulation, on PyBullet continuous control tasks, and directly on three different real robots: a tendon-driven elastic robot, a quadruped and an RC car. The noise sampling interval of gSDE permits to have a compromise between performance and smoothness, which allows training directly on the real robots without loss of performance. The code is available at https://github.com/DLR-RM/stable-baselines3.

(2 min) Improving the robustness of neural nets in regression tasks is key to their application in multiple domains. Deep learning-based approaches aim to achieve this goal either by improving their prediction of specific values (i.e., point prediction), or by producing prediction intervals (PIs) that quantify uncertainty. We present PIVEN, a deep neural network for producing both a PI and a value prediction. Our loss function expresses the value prediction as a function of the upper and lower bounds, thus ensuring that it falls within the interval without increasing model complexity. Moreover, our approach makes no assumptions regarding data distribution within the PI, making its value prediction more effective for various real-world problems. Experiments and ablation tests on known benchmarks show that our approach produces tighter uncertainty bounds than the current state-of-the-art approaches for producing PIs, while maintaining comparable performance to the state-of-the-art approach for value-prediction. Additionally, we go beyond previous work and include large image datasets in our evaluation, where PIVEN is combined with modern neural nets.

(2 min) We study the expressive power of deep ReLU neural networks for approximating functions in dilated shift-invariant spaces, which are widely used in signal processing, image processing, communications and so on. Approximation error bounds are estimated with respect to the width and depth of neural networks. The network construction is based on the bit extraction and data-fitting capacity of deep neural networks. As applications of our main results, the approximation rates of classical function spaces such as Sobolev spaces and Besov spaces are obtained. We also give lower bounds of the $L^p (1\le p \le \infty)$ approximation error for Sobolev spaces, which show that our construction of neural network is asymptotically optimal up to a logarithmic factor.

(2 min) To alleviate the data requirement for training effective binary classifiers in binary classification, many weakly supervised learning settings have been proposed. Among them, some consider using pairwise but not pointwise labels, when pointwise labels are not accessible due to privacy, confidentiality, or security reasons. However, as a pairwise label denotes whether or not two data points share a pointwise label, it cannot be easily collected if either point is equally likely to be positive or negative. Thus, in this paper, we propose a novel setting called pairwise comparison (Pcomp) classification, where we have only pairs of unlabeled data that we know one is more likely to be positive than the other. Firstly, we give a Pcomp data generation process, derive an unbiased risk estimator (URE) with theoretical guarantee, and further improve URE using correction functions. Secondly, we link Pcomp classification to noisy-label learning to develop a progressive URE and improve it by imposing consistency regularization. Finally, we demonstrate by experiments the effectiveness of our methods, which suggests Pcomp is a valuable and practically useful type of pairwise supervision besides the pairwise label.

(2 min) Recent advancements in deep learning have created many opportunities to solve real-world problems that remained unsolved for more than a decade. Automatic caption generation is a major research field, and the research community has done a lot of work on it in most common languages like English. Urdu is the national language of Pakistan and also much spoken and understood in the sub-continent region of Pakistan-India, and yet no work has been done for Urdu language caption generation. Our research aims to fill this gap by developing an attention-based deep learning model using techniques of sequence modeling specialized for the Urdu language. We have prepared a dataset in the Urdu language by translating a subset of the "Flickr8k" dataset containing 700 'man' images. We evaluate our proposed technique on this dataset and show that it can achieve a BLEU score of 0.83 in the Urdu language. We improve on the previous state-of-the-art by using better CNN architectures and optimization techniques. Furthermore, we provide a discussion on how the generated captions can be made correct grammar-wise.

(2 min) Ensembles of deep neural networks have achieved great success recently, but they do not offer a proper Bayesian justification. Moreover, while they allow for averaging of predictions over several hypotheses, they do not provide any guarantees for their diversity, leading to redundant solutions in function space. In contrast, particle-based inference methods, such as Stein variational gradient descent (SVGD), offer a Bayesian framework, but rely on the choice of a kernel to measure the similarity between ensemble members. In this work, we study different SVGD methods operating in the weight space, function space, and in a hybrid setting. % Defining the kernel directly on the neural network functions seems promising to overcome the limitations of deep ensembles. % However, ensuring diversity in function space while maintaining SVGD's theoretical guarantees is not trivial. % In this work, we provide an overview over different ensembling and SVGD methods in weight space and function space and propose new and assess their theoretical and empirical properties on synthetic and real-world tasks. We compare the SVGD approaches to other ensembling-based methods in terms of their theoretical properties and assess their empirical performance on synthetic and real-world tasks. We find that SVGD using functional and hybrid kernels can overcome the limitations of deep ensembles. It improves on functional diversity and uncertainty estimation and approaches the true Bayesian posterior more closely. Moreover, we show that using stochastic SVGD updates, as opposed to the standard deterministic ones, can further improve the performance.

(2 min) Scaling adaptive traffic-signal control involves dealing with combinatorial state and action spaces. Multi-agent reinforcement learning attempts to address this challenge by distributing control to specialized agents. However, specialization hinders generalization and transferability, and the computational graphs underlying neural-networks architectures -- dominating in the multi-agent setting -- do not offer the flexibility to handle an arbitrary number of entities which changes both between road networks, and over time as vehicles traverse the network. We introduce Inductive Graph Reinforcement Learning (IG-RL) based on graph-convolutional networks which adapts to the structure of any road network, to learn detailed representations of traffic-controllers and their surroundings. Our decentralized approach enables learning of a transferable-adaptive-traffic-signal-control policy. After being trained on an arbitrary set of road networks, our model can generalize to new road networks, traffic distributions, and traffic regimes, with no additional training and a constant number of parameters, enabling greater scalability compared to prior methods. Furthermore, our approach can exploit the granularity of available data by capturing the (dynamic) demand at both the lane and the vehicle levels. The proposed method is tested on both road networks and traffic settings never experienced during training. We compare IG-RL to multi-agent reinforcement learning and domain-specific baselines. In both synthetic road networks and in a larger experiment involving the control of the 3,971 traffic signals of Manhattan, we show that different instantiations of IG-RL outperform baselines.

(2 min) Federated Learning (FL) refers to the paradigm where multiple worker nodes (WNs) build a joint model by using local data. Despite extensive research, for a generic non-convex FL problem, it is not clear, how to choose the WNs' and the server's update directions, the minibatch sizes, and the local update frequency, so that the WNs use the minimum number of samples and communication rounds to achieve the desired solution. This work addresses the above question and considers a class of stochastic algorithms where the WNs perform a few local updates before communication. We show that when both the WN's and the server's directions are chosen based on a stochastic momentum estimator, the algorithm requires $\tilde{\mathcal{O}}(\epsilon^{-3/2})$ samples and $\tilde{\mathcal{O}}(\epsilon^{-1})$ communication rounds to compute an $\epsilon$-stationary solution. To the best of our knowledge, this is the first FL algorithm that achieves such {\it near-optimal} sample and communication complexities simultaneously. Further, we show that there is a trade-off curve between local update frequencies and local minibatch sizes, on which the above sample and communication complexities can be maintained. Finally, we show that for the classical FedAvg (a.k.a. Local SGD, which is a momentum-less special case of the STEM), a similar trade-off curve exists, albeit with worse sample and communication complexities. Our insights on this trade-off provides guidelines for choosing the four important design elements for FL algorithms, the update frequency, directions, and minibatch sizes to achieve the best performance.

(2 min) Several practical applications of reinforcement learning involve an agent learning from past data without the possibility of further exploration. Often these applications require us to 1) identify a near optimal policy or to 2) estimate the value of a target policy. For both tasks we derive \emph{exponential} information-theoretic lower bounds in discounted infinite horizon MDPs with a linear function representation for the action value function even if 1) \emph{realizability} holds, 2) the batch algorithm observes the exact reward and transition \emph{functions}, and 3) the batch algorithm is given the \emph{best} a priori data distribution for the problem class. Our work introduces a new `oracle + batch algorithm' framework to prove lower bounds that hold for every distribution. The work shows an exponential separation between batch and online reinforcement learning.

(2 min) Natural reading orders of words are crucial for information extraction from form-like documents. Despite recent advances in Graph Convolutional Networks (GCNs) on modeling spatial layout patterns of documents, they have limited ability to capture reading orders of given word-level node representations in a graph. We propose Reading Order Equivariant Positional Encoding (ROPE), a new positional encoding technique designed to apprehend the sequential presentation of words in documents. ROPE generates unique reading order codes for neighboring words relative to the target word given a word-level graph connectivity. We study two fundamental document entity extraction tasks including word labeling and word grouping on the public FUNSD dataset and a large-scale payment dataset. We show that ROPE consistently improves existing GCNs with a margin up to 8.4% F1-score.

(2 min) Developing efficient and principled neural architecture optimization methods is a critical challenge of modern deep learning. Recently, Liu et al.[19] proposed a splitting steepest descent (S2D) method that jointly optimizes the neural parameters and architectures based on progressively growing network structures by splitting neurons into multiple copies in a steepest descent fashion. However, S2D suffers from a local optimality issue when all the neurons become "splitting stable", a concept akin to local stability in parametric optimization. In this work, we develop a significant and surprising extension of the splitting descent framework that addresses the local optimality issue. The idea is to observe that the original S2D is unnecessarily restricted to splitting neurons into positive weighted copies. By simply allowing both positive and negative weights during splitting, we can eliminate the appearance of splitting stability in S2D and hence escape the local optima to obtain better performance. By incorporating signed splittings, we significantly extend the optimization power of splitting steepest descent both theoretically and empirically. We verify our method on various challenging benchmarks such as CIFAR-100, ImageNet and ModelNet40, on which we outperform S2D and other advanced methods on learning accurate and energy-efficient neural networks.

(2 min) We propose multirate training of neural networks: partitioning neural network parameters into "fast" and "slow" parts which are trained simultaneously using different learning rates. By choosing appropriate partitionings we can obtain large computational speed-ups for transfer learning tasks. We show that for various transfer learning applications in vision and NLP we can fine-tune deep neural networks in almost half the time, without reducing the generalization performance of the resulting model. We also discuss other splitting choices for the neural network parameters which are beneficial in enhancing generalization performance in settings where neural networks are trained from scratch. Finally, we propose an additional multirate technique which can learn different features present in the data by training the full network on different time scales simultaneously. The benefits of using this approach are illustrated for ResNet architectures on image data. Our paper unlocks the potential of using multirate techniques for neural network training and provides many starting points for future work in this area.

(2 min) Neural networks are known to suffer from catastrophic forgetting when trained on sequential datasets. While there have been numerous attempts to solve this problem in large-scale supervised classification, little has been done to overcome catastrophic forgetting in few-shot classification problems. We demonstrate that the popular gradient-based model-agnostic meta-learning algorithm (MAML) indeed suffers from catastrophic forgetting and introduce a Bayesian online meta-learning framework that tackles this problem. Our framework utilises Bayesian online learning and meta-learning along with Laplace approximation and variational inference to overcome catastrophic forgetting in few-shot classification problems. The experimental evaluations demonstrate that our framework can effectively achieve this goal in comparison with various baselines. As an additional utility, we also demonstrate empirically that our framework is capable of meta-learning on sequentially arriving few-shot tasks from a stationary task distribution.

(2 min) Deep compressed sensing assumes the data has sparse representation in a latent space, i.e., it is intrinsically of low-dimension. The original data is assumed to be mapped from a low-dimensional space through a low-to-high-dimensional generator. In this work, we propound how to design such a low-to-high dimensional deep learning-based generator suiting for compressed sensing, while satisfying robustness to universal adversarial perturbations in the latent domain. We also justify why the noise is considered in the latent space. The work is also buttressed with theoretical analysis on the robustness of the trained generator to adversarial perturbations. Experiments on real-world datasets are provided to substantiate the efficacy of the proposed \emph{generative model adversarial training for deep compressed sensing.}

(2 min) Based on a large dataset containing thousands of real-world networks ranging from genetic, protein interaction, and metabolic networks to brain, language, ecology, and social networks we search for defining structural measures of the different complex network domains (CND). We calculate 208 measures for all networks and using a comprehensive and scrupulous workflow of statistical and machine learning methods we investigated the limitations and possibilities of identifying the key graph measures of CNDs. Our approach managed to identify well distinguishable groups of network domains and confer their relevant features. These features turn out to be CND specific and not unique even at the level of individual CNDs. The presented methodology may be applied to other similar scenarios involving highly unbalanced and skewed datasets.

(2 min) In recent decades, Machine Learning (ML) has become extremely important for many computing applications. The pervasiveness of ultra-low-power embedded devices such as ESP32 or ESP32 Cam with tiny Machine Learning (tinyML) applications will enable the mass proliferation of Artificial Intelligent powered Embedded IoT Devices. In the last few years, the microcontroller device (Espressif ESP32) became powerful enough to be used for small/tiny machine learning (tinyML) tasks. The ease of use of platforms like Arduino IDE, MicroPython and TensorFlow Lite (TF) with tinyML application make it an indispensable topic of research for mobile robotics, modern computer science and electrical engineering. The goal of this paper is to analyze the speed of the Xtensa dual core 32-bit LX6 microprocessor by running a neural network application. The different number of inputs (9, 36, 144 and 576) inputted through the different number of neurons in neural networks with one and two hidden layers. Xtensa LX6 microprocessor has been analyzed because it comes inside with Espressif ESP32 and ESP32 Cam which are very easy to use, plug and play IoT device. In this paper speed of the Xtensa LX6 microprocessor in feed-forward mode has been analyzed.

(2 min) This paper makes two scientific contributions to the field of exoskeleton-based action and movement recognition. First, it presents a novel machine learning and pattern recognition-based framework that can detect a wide range of actions and movements - walking, walking upstairs, walking downstairs, sitting, standing, lying, stand to sit, sit to stand, sit to lie, lie to sit, stand to lie, and lie to stand, with an overall accuracy of 82.63%. Second, it presents a comprehensive comparative study of different learning approaches - Random Forest, Artificial Neural Network, Decision Tree, Multiway Decision Tree, Support Vector Machine, k-NN, Gradient Boosted Trees, Decision Stump, Auto MLP, Linear Regression, Vector Linear Regression, Random Tree, Na\"ive Bayes, Na\"ive Bayes (Kernel), Linear Discriminant Analysis, Quadratic Discriminant Analysis, and Deep Learning applied to this framework. The performance of each of these learning approaches was boosted by using the AdaBoost algorithm, and the Cross Validation approach was used for training and testing. The results show that in boosted form, the k- NN classifier outperforms all the other boosted learning approaches and is, therefore, the optimal learning method for this purpose. The results presented and discussed uphold the importance of this work to contribute towards augmenting the abilities of exoskeleton-based assisted and independent living of the elderly in the future of Internet of Things-based living environments, such as Smart Homes. As a specific use case, we also discuss how the findings of our work are relevant for augmenting the capabilities of the Hybrid Assistive Limb exoskeleton, a highly functional lower limb exoskeleton.

(2 min) Massive access is a critical design challenge of Internet of Things (IoT) networks. In this paper, we consider the grant-free uplink transmission of an IoT network with a multiple-antenna base station (BS) and a large number of single-antenna IoT devices. Taking into account the sporadic nature of IoT devices, we formulate the joint activity detection and channel estimation (JADCE) problem as a group-sparse matrix estimation problem. This problem can be solved by applying the existing compressed sensing techniques, which however either suffer from high computational complexities or lack of algorithm robustness. To this end, we propose a novel algorithm unrolling framework based on the deep neural network to simultaneously achieve low computational complexity and high robustness for solving the JADCE problem. Specifically, we map the original iterative shrinkage thresholding algorithm (ISTA) into an unrolled recurrent neural network (RNN), thereby improving the convergence rate and computational efficiency through end-to-end training. Moreover, the proposed algorithm unrolling approach inherits the structure and domain knowledge of the ISTA, thereby maintaining the algorithm robustness, which can handle non-Gaussian preamble sequence matrix in massive access. With rigorous theoretical analysis, we further simplify the unrolled network structure by reducing the redundant training parameters. Furthermore, we prove that the simplified unrolled deep neural network structures enjoy a linear convergence rate. Extensive simulations based on various preamble signatures show that the proposed unrolled networks outperform the existing methods in terms of the convergence rate, robustness and estimation accuracy.

(2 min) Deep visual models are susceptible to adversarial perturbations to inputs. Although these signals are carefully crafted, they still appear noise-like patterns to humans. This observation has led to the argument that deep visual representation is misaligned with human perception. We counter-argue by providing evidence of human-meaningful patterns in adversarial perturbations. We first propose an attack that fools a network to confuse a whole category of objects (source class) with a target label. Our attack also limits the unintended fooling by samples from non-sources classes, thereby circumscribing human-defined semantic notions for network fooling. We show that the proposed attack not only leads to the emergence of regular geometric patterns in the perturbations, but also reveals insightful information about the decision boundaries of deep models. Exploring this phenomenon further, we alter the `adversarial' objective of our attack to use it as a tool to `explain' deep visual representation. We show that by careful channeling and projection of the perturbations computed by our method, we can visualize a model's understanding of human-defined semantic notions. Finally, we exploit the explanability properties of our perturbations to perform image generation, inpainting and interactive image manipulation by attacking adversarialy robust `classifiers'.In all, our major contribution is a novel pragmatic adversarial attack that is subsequently transformed into a tool to interpret the visual models. The article also makes secondary contributions in terms of establishing the utility of our attack beyond the adversarial objective with multiple interesting applications.

(2 min) Nowadays, live video streaming events have become a mainstay in viewer's communication in large international enterprises. Provided that viewers are distributed worldwide, the main challenge resides on how to schedule the optimal event's time so as to improve both the viewer's engagement and adoption. In this paper we present a multi-task deep reinforcement learning model to select the time of a live video streaming event, aiming to optimize the viewer's engagement and adoption at the same time. We consider the engagement and adoption of the viewers as independent tasks and formulate a unified loss function to learn a common policy. In addition, we account for the fact that each task might have different contribution to the training strategy of the agent. Therefore, to determine the contribution of each task to the agent's training, we design a Transformer's architecture for the state-action transitions of each task. We evaluate our proposed model on four real-world datasets, generated by the live video streaming events of four large enterprises spanning from January 2019 until March 2021. Our experiments demonstrate the effectiveness of the proposed model when compared with several state-of-the-art strategies. For reproduction purposes, our evaluation datasets and implementation are publicly available at https://github.com/stefanosantaris/merlin.

(2 min) Repeated use of a data sample via adaptively chosen queries can rapidly lead to overfitting, wherein the issued queries yield answers on the sample that differ wildly from the values of those queries on the underlying data distribution. Differential privacy provides a tool to ensure generalization despite adaptively-chosen queries, but its worst-case nature means that it cannot, for example, yield improved results for low-variance queries. In this paper, we give a simple new characterization that illuminates the core problem of adaptive data analysis. We show explicitly that the harms of adaptivity come from the covariance between the behavior of future queries and a Bayes factor-based measure of how much information about the data sample was encoded in the responses given to past queries. We leverage this intuition to introduce a new stability notion; we then use it to prove new generalization results for the most basic noise-addition mechanisms (Laplace and Gaussian noise addition), with guarantees that scale with the variance of the queries rather than the square of their range. Our characterization opens the door to new insights and new algorithms for the fundamental problem of achieving generalization in adaptive data analysis.

(2 min) We propose TD-GEN, a graph generation framework based on tree decomposition, and introduce a reduced upper bound on the maximum number of decisions needed for graph generation. The framework includes a permutation invariant tree generation model which forms the backbone of graph generation. Tree nodes are supernodes, each representing a cluster of nodes in the graph. Graph nodes and edges are incrementally generated inside the clusters by traversing the tree supernodes, respecting the structure of the tree decomposition, and following node sharing decisions between the clusters. Finally, we discuss the shortcomings of standard evaluation criteria based on statistical properties of the generated graphs as performance measures. We propose to compare the performance of models based on likelihood. Empirical results on a variety of standard graph generation datasets demonstrate the superior performance of our method.

(2 min) The evaluation of rare but high-stakes events remains one of the main difficulties in obtaining reliable policies from intelligent agents, especially in large or continuous state/action spaces where limited scalability enforces the use of a prohibitively large number of testing iterations. On the other hand, a biased or inaccurate policy evaluation in a safety-critical system could potentially cause unexpected catastrophic failures during deployment. In this paper, we propose the Accelerated Policy Evaluation (APE) method, which simultaneously uncovers rare events and estimates the rare event probability in Markov decision processes. The APE method treats the environment nature as an adversarial agent and learns towards, through adaptive importance sampling, the zero-variance sampling distribution for the policy evaluation. Moreover, APE is scalable to large discrete or continuous spaces by incorporating function approximators. We investigate the convergence properties of proposed algorithms under suitable regularity conditions. Our empirical studies show that APE estimates rare event probability with a smaller variance while only using orders of magnitude fewer samples compared to baseline methods in both multi-agent and single-agent environments.

(2 min) We consider the offline reinforcement learning (RL) setting where the agent aims to optimize the policy solely from the data without further environment interactions. In offline RL, the distributional shift becomes the primary source of difficulty, which arises from the deviation of the target policy being optimized from the behavior policy used for data collection. This typically causes overestimation of action values, which poses severe problems for model-free algorithms that use bootstrapping. To mitigate the problem, prior offline RL algorithms often used sophisticated techniques that encourage underestimation of action values, which introduces an additional set of hyperparameters that need to be tuned properly. In this paper, we present an offline RL algorithm that prevents overestimation in a more principled way. Our algorithm, OptiDICE, directly estimates the stationary distribution corrections of the optimal policy and does not rely on policy-gradients, unlike previous offline RL algorithms. Using an extensive set of benchmark datasets for offline RL, we show that OptiDICE performs competitively with the state-of-the-art methods.

(2 min) The mean squared error loss is widely used in many applications, including auto-encoders, multi-target regression, and matrix factorization, to name a few. Despite computational advantages due to its differentiability, it is not robust to outliers. In contrast, l_p norms are known to be robust, but cannot be optimized via, e.g., stochastic gradient descent, as they are non-differentiable. We propose an algorithm inspired by so-called model-based optimization (MBO) [35, 36], which replaces a non-convex objective with a convex model function and alternates between optimizing the model function and updating the solution. We apply this to robust regression, proposing SADM, a stochastic variant of the Online Alternating Direction Method of Multipliers (OADM) [50] to solve the inner optimization in MBO. We show that SADM converges with the rate O(log T/T). Finally, we demonstrate experimentally (a) the robustness of l_p norms to outliers and (b) the efficiency of our proposed model-based algorithms in comparison with gradient methods on autoencoders and multi-target regression.

(2 min) Since most industrial control applications use PID controllers, PID tuning and anti-windup measures are significant problems. This paper investigates tuning the feedback gains of a PID controller via back-calculation and automatic differentiation tools. In particular, we episodically use a cost function to generate gradients and perform gradient descent to improve controller performance. We provide a theoretical framework for analyzing this non-convex optimization and establish a relationship between back-calculation and disturbance feedback policies. We include numerical experiments on linear systems with actuator saturation to show the efficacy of this approach.

(2 min) We describe some recent approaches to score-based explanations for query answers in databases and outcomes from classification models in machine learning. The focus is on work done by the author and collaborators. Special emphasis is placed on declarative approaches based on answer-set programming to the use of counterfactual reasoning for score specification and computation. Several examples that illustrate the flexibility of these methods are shown.

(2 min) Existing computer vision research in categorization struggles with fine-grained attributes recognition due to the inherently high intra-class variances and low inter-class variances. SOTA methods tackle this challenge by locating the most informative image regions and rely on them to classify the complete image. The most recent work, Vision Transformer (ViT), shows its strong performance in both traditional and fine-grained classification tasks. In this work, we propose a multi-stage ViT framework for fine-grained image classification tasks, which localizes the informative image regions without requiring architectural changes using the inherent multi-head self-attention mechanism. We also introduce attention-guided augmentations for improving the model's capabilities. We demonstrate the value of our approach by experimenting with four popular fine-grained benchmarks: CUB-200-2011, Stanford Cars, Stanford Dogs, and FGVC7 Plant Pathology. We also prove our model's interpretability via qualitative results.

(2 min) Variational Monte Carlo (VMC) is an approach for computing ground-state wavefunctions that has recently become more powerful due to the introduction of neural network-based wavefunction parametrizations. However, efficiently training neural wavefunctions to converge to an energy minimum remains a difficult problem. In this work, we analyze optimization and sampling methods used in VMC and introduce alterations to improve their performance. First, based on theoretical convergence analysis in a noiseless setting, we motivate a new optimizer that we call the Rayleigh-Gauss-Newton method, which can improve upon gradient descent and natural gradient descent to achieve superlinear convergence. Second, in order to realize this favorable comparison in the presence of stochastic noise, we analyze the effect of sampling error on VMC parameter updates and experimentally demonstrate that it can be reduced by the parallel tempering method. In particular, we demonstrate that RGN can be made robust to energy spikes that occur when new regions of configuration space become available to the sampler over the course of optimization. Finally, putting theory into practice, we apply our enhanced optimization and sampling methods to the transverse-field Ising and XXZ models on large lattices, yielding ground-state energy estimates with remarkably high accuracy after just 200-500 parameter updates.

(2 min) Transferability estimation is an essential problem in transfer learning to predict how good the performance is when transfer a source model (source task) to a target task. Recent analytical transferability metrics have been widely used for source model selection and multi-task learning. Earlier metrics does not work sufficiently well under the challenging cross-domain cross-task transfer settings, but recent OTCE score achieves a noteworthy performance using auxiliary tasks. A simplified version named OT-based NCE score sacrifices accuracy to be more efficient, but it can be further improved. Consequently, we propose a practical transferability metric called JC-NCE score to further improve the cross-domain cross-task transferability estimation performance, which is more efficient than the OTCE score and more accurate than the OT-based NCE score. Specifically, we build the joint correspondences between source and target data via solving an optimal transport problem with considering both the sample distance and label distance, and then compute the transferability score as the negative conditional entropy. Extensive validations under the intra-dataset and inter-dataset transfer settings demonstrate that our JC-NCE score outperforms the OT-based NCE score with about 7% and 12% gains, respectively.

(2 min) A neural network is essentially a high-dimensional complex mapping model by adjusting network weights for feature fitting. However, the spectral bias in network training leads to unbearable training epochs for fitting the high-frequency components in broadband signals. To improve the fitting efficiency of high-frequency components, the PhaseDNN was proposed recently by combining complex frequency band extraction and frequency shift techniques [Cai et al. SIAM J. SCI. COMPUT. 42, A3285 (2020)]. Our paper is devoted to an alternative candidate for fitting complex signals with high-frequency components. Here, a parallel frequency function-deep neural network (PFF-DNN) is proposed to suppress computational overhead while ensuring fitting accuracy by utilizing fast Fourier analysis of broadband signals and the spectral bias nature of neural networks. The effectiveness and efficiency of the proposed PFF-DNN method are verified based on detailed numerical experiments for six typical broadband signals.

(2 min) Blindly decoding a signal requires estimating its unknown transmit parameters, compensating for the wireless channel impairments, and identifying the modulation type. While deep learning can solve complex problems, digital signal processing (DSP) is interpretable and can be more computationally efficient. To combine both, we propose the dual path network (DPN). It consists of a signal path of DSP operations that recover the signal, and a feature path of neural networks that estimate the unknown transmit parameters. By interconnecting the paths over several recovery stages, later stages benefit from the recovered signals and reuse all the previously extracted features. The proposed design is demonstrated to provide 5% improvement in modulation classification compared to alternative designs lacking either feature sharing or access to recovered signals. The estimation results of DPN along with its blind decoding performance are shown to outperform a blind signal processing algorithm for BPSK and QPSK on a simulated dataset. An over-the-air software-defined-radio capture was used to verify DPN results at high SNRs. DPN design can process variable length inputs and is shown to outperform relying on fixed length inputs with prediction averaging on longer signals by up to 15% in modulation classification.

(2 min) The utilization of computer technology to solve problems in medical scenarios has attracted considerable attention in recent years, which still has great potential and space for exploration. Among them, machine learning has been widely used in the prediction, diagnosis and even treatment of Sepsis. However, state-of-the-art methods require large amounts of labeled medical data for supervised learning. In real-world applications, the lack of labeled data will cause enormous obstacles if one hospital wants to deploy a new Sepsis detection system. Different from the supervised learning setting, we need to use known information (e.g., from another hospital with rich labeled data) to help build a model with acceptable performance, i.e., transfer learning. In this paper, we propose a semi-supervised optimal transport with self-paced ensemble framework for Sepsis early detection, called SPSSOT, to transfer knowledge from the other that has rich labeled data. In SPSSOT, we first extract the same clinical indicators from the source domain (e.g., hospital with rich labeled data) and the target domain (e.g., hospital with little labeled data), then we combine the semi-supervised domain adaptation based on optimal transport theory with self-paced under-sampling to avoid a negative transfer possibly caused by covariate shift and class imbalance. On the whole, SPSSOT is an end-to-end transfer learning method for Sepsis early detection which can automatically select suitable samples from two domains respectively according to the number of iterations and align feature space of two domains. Extensive experiments on two open clinical datasets demonstrate that comparing with other methods, our proposed SPSSOT, can significantly improve the AUC values with only 1% labeled data in the target domain in two transfer learning scenarios, MIMIC $rightarrow$ Challenge and Challenge $rightarrow$ MIMIC.

(2 min) In recent years particle filters have being used as components in systems optimized end-to-end with gradient descent. However, the resampling step in a particle filter is not differentiable, which biases gradients and interferes with optimization. To remedy this problem, several differentiable variants of resampling have been proposed, all of which modify the behavior of the particle filter in significant and potentially undesirable ways. In this paper, we show how to obtain unbiased estimators of the gradient of the marginal likelihood by only modifying messages used in backpropagation, leaving the standard forward pass of a particle filter unchanged. Our method is simple to implement, has a low computational overhead, does not introduce additional hyperparameters, and extends to derivatives of higher orders. We call it stop-gradient resampling, since it can easily be implemented with automatic differentiation libraries using the stop-gradient operator instead of explicitly modifying the backward messages.

(2 min) In this paper, we propose a max-min entropy framework for reinforcement learning (RL) to overcome the limitation of the maximum entropy RL framework in model-free sample-based learning. Whereas the maximum entropy RL framework guides learning for policies to reach states with high entropy in the future, the proposed max-min entropy framework aims to learn to visit states with low entropy and maximize the entropy of these low-entropy states to promote exploration. For general Markov decision processes (MDPs), an efficient algorithm is constructed under the proposed max-min entropy framework based on disentanglement of exploration and exploitation. Numerical results show that the proposed algorithm yields drastic performance improvement over the current state-of-the-art RL algorithms.

(2 min) Gradient-based meta-learning and hyperparameter optimization have seen significant progress recently, enabling practical end-to-end training of neural networks together with many hyperparameters. Nevertheless, existing approaches are relatively expensive as they need to compute second-order derivatives and store a longer computational graph. This cost prevents scaling them to larger network architectures. We present EvoGrad, a new approach to meta-learning that draws upon evolutionary techniques to more efficiently compute hypergradients. EvoGrad estimates hypergradient with respect to hyperparameters without calculating second-order gradients, or storing a longer computational graph, leading to significant improvements in efficiency. We evaluate EvoGrad on two substantial recent meta-learning applications, namely cross-domain few-shot learning with feature-wise transformations and noisy label learning with MetaWeightNet. The results show that EvoGrad significantly improves efficiency and enables scaling meta-learning to bigger CNN architectures such as from ResNet18 to ResNet34.

(2 min) In this paper, we present an algorithm to efficiently learn socially-compliant navigation policies from observations of human trajectories. As mobile robots come to inhabit and traffic social spaces, they must account for social cues and behave in a socially compliant manner. We focus on learning such cues from examples. We describe an inverse reinforcement learning based algorithm which learns from human trajectory observations without knowing their specific actions. We increase the sample-efficiency of our approach over alternative methods by leveraging the notion of a replay buffer (found in many off-policy reinforcement learning methods) to eliminate the additional sample complexity associated with inverse reinforcement learning. We evaluate our method by training agents using publicly available pedestrian motion data sets and compare it to related methods. We show that our approach yields better performance while also decreasing training time and sample complexity.

(2 min) We advocate for a practical Maximum Likelihood Estimation (MLE) approach for regression and forecasting, as an alternative to the typical approach of Empirical Risk Minimization (ERM) for a specific target metric. This approach is better suited to capture inductive biases such as prior domain knowledge in datasets, and can output post-hoc estimators at inference time that can optimize different types of target metrics. We present theoretical results to demonstrate that our approach is always competitive with any estimator for the target metric under some general conditions, and in many practical settings (such as Poisson Regression) can actually be much superior to ERM. We demonstrate empirically that our method instantiated with a well-designed general purpose mixture likelihood family can obtain superior performance over ERM for a variety of tasks across time-series forecasting and regression datasets with different data distributions.

(2 min) There has recently been significant interest in training reinforcement learning (RL) agents in vision-based environments. This poses many challenges, such as high dimensionality and potential for observational overfitting through spurious correlations. A promising approach to solve both of these problems is a self-attention bottleneck, which provides a simple and effective framework for learning high performing policies, even in the presence of distractions. However, due to poor scalability of attention architectures, these methods do not scale beyond low resolution visual inputs, using large patches (thus small attention matrices). In this paper we make use of new efficient attention algorithms, recently shown to be highly effective for Transformers, and demonstrate that these new techniques can be applied in the RL setting. This allows our attention-based controllers to scale to larger visual inputs, and facilitate the use of smaller patches, even individual pixels, improving generalization. In addition, we propose a new efficient algorithm approximating softmax attention with what we call hybrid random features, leveraging the theory of angular kernels. We show theoretically and empirically that hybrid random features is a promising approach when using attention for vision-based RL.

(2 min) \textit{Differentiable ARchiTecture Search} (DARTS) has recently become the mainstream of neural architecture search (NAS) due to its efficiency and simplicity. With a gradient-based bi-level optimization, DARTS alternately optimizes the inner model weights and the outer architecture parameter in a weight-sharing supernet. A key challenge to the scalability and quality of the learned architectures is the need for differentiating through the inner-loop optimisation. While much has been discussed about several potentially fatal factors in DARTS, the architecture gradient, a.k.a. hypergradient, has received less attention. In this paper, we tackle the hypergradient computation in DARTS based on the implicit function theorem, making it only depends on the obtained solution to the inner-loop optimization and agnostic to the optimization path. To further reduce the computational requirements, we formulate a stochastic hypergradient approximation for differentiable NAS, and theoretically show that the architecture optimization with the proposed method, named iDARTS, is expected to converge to a stationary point. Comprehensive experiments on two NAS benchmark search spaces and the common NAS search space verify the effectiveness of our proposed method. It leads to architectures outperforming, with large margins, those learned by the baseline methods.

(2 min) We introduce OLIVAW, an AI Othello player adopting the design principles of the famous AlphaGo series. The main motivation behind OLIVAW was to attain exceptional competence in a non-trivial board game at a tiny fraction of the cost of its illustrious predecessors. In this paper, we show how the AlphaGo Zero's paradigm can be successfully applied to the popular game of Othello using only commodity hardware and free cloud services. While being simpler than Chess or Go, Othello maintains a considerable search space and difficulty in evaluating board positions. To achieve this result, OLIVAW implements some improvements inspired by recent works to accelerate the standard AlphaGo Zero learning process. The main modification implies doubling the positions collected per game during the training phase, by including also positions not played but largely explored by the agent. We tested the strength of OLIVAW in three different ways: by pitting it against Edax, the strongest open-source Othello engine, by playing anonymous games on the web platform OthelloQuest, and finally in two in-person matches against top-notch human players: a national champion and a former world champion.

(2 min) We develop a variational framework to understand the properties of functions learned by deep neural networks with ReLU activation functions fit to data. We propose a new function space, which is reminiscent of classical bounded variation spaces, that captures the compositional structure associated with deep neural networks. We derive a representer theorem showing that deep ReLU networks are solutions to regularized data fitting problems in this function space. The function space consists of compositions of functions from the (non-reflexive) Banach spaces of second-order bounded variation in the Radon domain. These are Banach spaces with sparsity-promoting norms, giving insight into the role of sparsity in deep neural networks. The neural network solutions have skip connections and rank bounded weight matrices, providing new theoretical support for these common architectural choices. The variational problem we study can be recast as a finite-dimensional neural network training problem with regularization schemes related to the notions of weight decay and path-norm regularization. Finally, our analysis builds on techniques from variational spline theory, providing new connections between deep neural networks and splines.

(2 min) In this paper, we explore contrastive learning for few-shot classification, in which we propose to use it as an additional auxiliary training objective acting as a data-dependent regularizer to promote more general and transferable features. In particular, we present a novel attention-based spatial contrastive objective to learn locally discriminative and class-agnostic features. As a result, our approach overcomes some of the limitations of the cross-entropy loss, such as its excessive discrimination towards seen classes, which reduces the transferability of features to unseen classes. With extensive experiments, we show that the proposed method outperforms state-of-the-art approaches, confirming the importance of learning good and transferable embeddings for few-shot learning.

(2 min) Biological neurons have adaptive nature and perform complex computations involving the filtering of redundant information. However, most common neural cell models, including biologically plausible, such as Hodgkin-Huxley or Izhikevich, do not possess predictive dynamics on a single-cell level. Moreover, the modern rules of synaptic plasticity or interconnections weights adaptation also do not provide grounding for the ability of neurons to adapt to the ever-changing input signal intensity. While natural neuron synaptic growth is precisely controlled and restricted by protein supply and recycling, weight correction rules such as widely used STDP are efficiently unlimited in change rate and scale. The present article introduces new mechanics of interconnection between neuron firing rate homeostasis and weight change through STDP growth bounded by abstract protein reserve, controlled by the intracellular optimization algorithm. We show how these cellular dynamics help neurons filter out the intense noise signals to help neurons keep a stable firing rate. We also examine that such filtering does not affect the ability of neurons to recognize the correlated inputs in unsupervised mode. Such an approach might be used in the machine learning domain to improve the robustness of AI systems.

(2 min) We study defense strategies against reward poisoning attacks in reinforcement learning. As a threat model, we consider attacks that minimally alter rewards to make the attacker's target policy uniquely optimal under the poisoned rewards, with the optimality gap specified by an attack parameter. Our goal is to design agents that are robust against such attacks in terms of the worst-case utility w.r.t. the true, unpoisoned, rewards while computing their policies under the poisoned rewards. We propose an optimization framework for deriving optimal defense policies, both when the attack parameter is known and unknown. Moreover, we show that defense policies that are solutions to the proposed optimization problems have provable performance guarantees. In particular, we provide the following bounds with respect to the true, unpoisoned, rewards: a) lower bounds on the expected return of the defense policies, and b) upper bounds on how suboptimal these defense policies are compared to the attacker's target policy. We conclude the paper by illustrating the intuitions behind our formal results, and showing that the derived bounds are non-trivial.

(2 min) We propose a method for efficiently incorporating constraints into a stochastic gradient Langevin framework for the training of deep neural networks. Constraints allow direct control of the parameter space of the model. Appropriately designed, they reduce the vanishing/exploding gradient problem, control weight magnitudes and stabilize deep neural networks and thus improve the robustness of training algorithms and the generalization capabilities of the trained neural network. We present examples of constrained training methods motivated by orthogonality preservation for weight matrices and explicit weight normalizations. We describe the methods in the overdamped formulation of Langevin dynamics and the underdamped form, in which momenta help to improve sampling efficiency. The methods are explored in test examples in image classification and natural language processing.

(2 min) The adversarial machine learning literature is largely partitioned into evasion attacks on testing data and poisoning attacks on training data. In this work, we show that adversarial examples, originally intended for attacking pre-trained models, are even more effective for data poisoning than recent methods designed specifically for poisoning. Our findings indicate that adversarial examples, when assigned the original label of their natural base image, cannot be used to train a classifier for natural images. Furthermore, when adversarial examples are assigned their adversarial class label, they are useful for training. This suggests that adversarial examples contain useful semantic content, just with the ``wrong'' labels (according to a network, but not a human). Our method, adversarial poisoning, is substantially more effective than existing poisoning methods for secure dataset release, and we release a poisoned version of ImageNet, ImageNet-P, to encourage research into the strength of this form of data obfuscation.

(2 min) Most data is automatically collected and only ever "seen" by algorithms. Yet, data compressors preserve perceptual fidelity rather than just the information needed by algorithms performing downstream tasks. In this paper, we characterize the bit-rate required to ensure high performance on all predictive tasks that are invariant under a set of transformations, such as data augmentations. Based on our theory, we design unsupervised objectives for training neural compressors. Using these objectives, we train a generic image compressor that achieves substantial rate savings (more than $1000\times$ on ImageNet) compared to JPEG on 8 datasets, without decreasing downstream classification performance.

(2 min) We propose a numerical accountant for evaluating the tight $(\varepsilon,\delta)$-privacy loss for algorithms with discrete one dimensional output. The method is based on the privacy loss distribution formalism and it uses the recently introduced fast Fourier transform based accounting technique. We carry out an error analysis of the method in terms of moment bounds of the privacy loss distribution which leads to rigorous lower and upper bounds for the true $(\varepsilon,\delta)$-values. As an application, we present a novel approach to accurate privacy accounting of the subsampled Gaussian mechanism. This completes the previously proposed analysis by giving strict lower and upper bounds for the privacy parameters. We demonstrate the performance of the accountant on the binomial mechanism and show that our approach allows decreasing noise variance up to 75 percent at equal privacy compared to existing bounds in the literature. We also illustrate how to compute tight bounds for the exponential mechanism applied to counting queries.

(2 min) Metric learning is a widely used method for few shot learning in which the quality of prototypes plays a key role in the algorithm. In this paper we propose the trainable prototypes for distance measure instead of the artificial ones within the meta-training and task-training framework. Also to avoid the disadvantages that the episodic meta-training brought, we adopt non-episodic meta-training based on self-supervised learning. Overall we solve the few-shot tasks in two phases: meta-training a transferable feature extractor via self-supervised learning and training the prototypes for metric classification. In addition, the simple attention mechanism is used in both meta-training and task-training. Our method achieves state-of-the-art performance in a variety of established few-shot tasks on the standard few-shot visual classification dataset, with about 20% increase compared to the available unsupervised few-shot learning methods.

(2 min) We show that the influence of a subset of the training samples can be removed -- or "forgotten" -- from the weights of a network trained on large-scale image classification tasks, and we provide strong computable bounds on the amount of remaining information after forgetting. Inspired by real-world applications of forgetting techniques, we introduce a novel notion of forgetting in mixed-privacy setting, where we know that a "core" subset of the training samples does not need to be forgotten. While this variation of the problem is conceptually simple, we show that working in this setting significantly improves the accuracy and guarantees of forgetting methods applied to vision classification tasks. Moreover, our method allows efficient removal of all information contained in non-core data by simply setting to zero a subset of the weights with minimal loss in performance. We achieve these results by replacing a standard deep network with a suitable linear approximation. With opportune changes to the network architecture and training procedure, we show that such linear approximation achieves comparable performance to the original network and that the forgetting problem becomes quadratic and can be solved efficiently even for large models. Unlike previous forgetting methods on deep networks, ours can achieve close to the state-of-the-art accuracy on large scale vision tasks. In particular, we show that our method allows forgetting without having to trade off the model accuracy.

(2 min) Malware currently presents a number of serious threats to computer users. Signature-based malware detection methods are limited in detecting new malware samples that are significantly different from known ones. Therefore, machine learning-based methods have been proposed, but there are two challenges these methods face. The first is to model the full semantics behind the assembly code of malware. The second challenge is to provide interpretable results while keeping excellent detection performance. In this paper, we propose an Interpretable MAlware Detector (I-MAD) that outperforms state-of-the-art static malware detection models regarding accuracy with excellent interpretability. To improve the detection performance, I-MAD incorporates a novel network component called the Galaxy Transformer network that can understand assembly code at the basic block, function, and executable levels. It also incorporates our proposed interpretable feed-forward neural network to provide interpretations for its detection results by quantifying the impact of each feature with respect to the prediction. Experiment results show that our model significantly outperforms existing state-of-the-art static malware detection models and presents meaningful interpretations.

(2 min) This paper develops a new mathematical-statistical approach to analyze a class of Flajolet-Martin algorithms (FMa), and provides analytical confidence intervals for the number F0 of distinct elements in a stream, based on Chernoff bounds. The class of FMa has reached a significant popularity in bigdata stream learning, and the attention of the literature has mainly been based on algorithmic aspects, basically complexity optimality, while the statistical analysis of these class of algorithms has been often faced heuristically. The analysis provided here shows deep connections with mathematical special functions and with extreme value theory. The latter connection may help in explaining heuristic considerations, while the first opens many numerical issues, faced at the end of the present paper. Finally, the algorithms are tested on an anonymized real data stream and MonteCarlo simulations are provided to support our analytical choice in this context.

(2 min) Training images with data transformations have been suggested as contrastive examples to complement the testing set for generalization performance evaluation of deep neural networks (DNNs). In this work, we propose a practical framework ContRE (The word "contre" means "against" or "versus" in French.) that uses Contrastive examples for DNN geneRalization performance Estimation. Specifically, ContRE follows the assumption in contrastive learning that robust DNN models with good generalization performance are capable of extracting a consistent set of features and making consistent predictions from the same image under varying data transformations. Incorporating with a set of randomized strategies for well-designed data transformations over the training set, ContRE adopts classification errors and Fisher ratios on the generated contrastive examples to assess and analyze the generalization performance of deep models in complement with a testing set. To show the effectiveness and the efficiency of ContRE, extensive experiments have been done using various DNN models on three open source benchmark datasets with thorough ablation studies and applicability analyses. Our experiment results confirm that (1) behaviors of deep models on contrastive examples are strongly correlated to what on the testing set, and (2) ContRE is a robust measure of generalization performance complementing to the testing set in various settings.

(2 min) The recently proposed Fast Fourier Transform (FFT)-based accountant for evaluating $(\varepsilon,\delta)$-differential privacy guarantees using the privacy loss distribution formalism has been shown to give tighter bounds than commonly used methods such as R\'enyi accountants when applied to homogeneous compositions, i.e., to compositions of identical mechanisms. In this paper, we extend this approach to heterogeneous compositions. We carry out a full error analysis that allows choosing the parameters of the algorithm such that a desired accuracy is obtained. The analysis also extends previous results by taking into account all the parameters of the algorithm. Using the error analysis, we also give a bound for the computational complexity in terms of the error which is analogous to and slightly tightens the one given by Murtagh and Vadhan (2018). We also show how to speed up the evaluation of tight privacy guarantees using the Plancherel theorem at the cost of increased pre-computation and memory usage.

(2 min) Tidal range structures have been considered for large scale electricity generation for their potential ability to produce reasonable predictable energy without the emission of greenhouse gases. Once the main forcing components for driving the tides have deterministic dynamics, the available energy in a given tidal power plant has been estimated, through analytical and numerical optimisation routines, as a mostly predictable event. This constraint imposes state-of-art flexible operation methods to rely on tidal predictions (concurrent with measured data and up to a multiple of half-tidal cycles into the future) to infer best operational strategies for tidal lagoons, with the additional cost of requiring to run optimisation routines for every new tide. In this paper, we propose a novel optimised operation of tidal lagoons with proximal policy optimisation through Unity ML-Agents. We compare this technique with 6 different operation optimisation approaches (baselines) devised from the literature, utilising the Swansea Bay Tidal Lagoon as a case study. We show that our approach is successful in maximising energy generation through an optimised operational policy of turbines and sluices, yielding competitive results with state-of-the-art methods of optimisation, regardless of test data used, requiring training once and performing real-time flexible control with measured ocean data only.

(2 min) The uptake of machine learning (ML) approaches in the social and health sciences has been rather slow, and research using ML for social and health research questions remains fragmented. This may be due to the separate development of research in the computational/data versus social and health sciences as well as a lack of accessible overviews and adequate training in ML techniques for non data science researchers. This paper provides a meta-mapping of research questions in the social and health sciences to appropriate ML approaches, by incorporating the necessary requirements to statistical analysis in these disciplines. We map the established classification into description, prediction, and causal inference to common research goals, such as estimating prevalence of adverse health or social outcomes, predicting the risk of an event, and identifying risk factors or causes of adverse outcomes. This meta-mapping aims at overcoming disciplinary barriers and starting a fluid dialogue between researchers from the social and health sciences and methodologically trained researchers. Such mapping may also help to fully exploit the benefits of ML while considering domain-specific aspects relevant to the social and health sciences, and hopefully contribute to the acceleration of the uptake of ML applications to advance both basic and applied social and health sciences research.

(2 min) Acoustic Event Classification (AEC) has become a significant task for machines to perceive the surrounding auditory scene. However, extracting effective representations that capture the underlying characteristics of the acoustic events is still challenging. Previous methods mainly focused on designing the audio features in a `hand-crafted' manner. Interestingly, data-learnt features have been recently reported to show better performance. Up to now, these were only considered on the frame level. In this article, we propose an unsupervised learning framework to learn a vector representation of an audio sequence for AEC. This framework consists of a Recurrent Neural Network (RNN) encoder and an RNN decoder, which respectively transforms the variable-length audio sequence into a fixed-length vector and reconstructs the input sequence on the generated vector. After training the encoder-decoder, we feed the audio sequences to the encoder and then take the learnt vectors as the audio sequence representations. Compared with previous methods, the proposed method can not only deal with the problem of arbitrary-lengths of audio streams, but also learn the salient information of the sequence. Extensive evaluation on a large-size acoustic event database is performed, and the empirical results demonstrate that the learnt audio sequence representation yields a significant performance improvement by a large margin compared with other state-of-the-art hand-crafted sequence features for AEC.

(2 min) Over the past two decades, recommender systems have attracted a lot of interest due to the explosion in the amount of data in online applications. A particular attention has been paid to collaborative filtering, which is the most widely used in applications that involve information recommendations. Collaborative filtering (CF) uses the known preference of a group of users to make predictions and recommendations about the unknown preferences of other users (recommendations are made based on the past behavior of users). First introduced in the 1990s, a wide variety of increasingly successful models have been proposed. Due to the success of machine learning techniques in many areas, there has been a growing emphasis on the application of such algorithms in recommendation systems. In this article, we present an overview of the CF approaches for recommender systems, their two main categories, and their evaluation metrics. We focus on the application of classical Machine Learning algorithms to CF recommender systems by presenting their evolution from their first use-cases to advanced Machine Learning models. We attempt to provide a comprehensive and comparative overview of CF systems (with python implementations) that can serve as a guideline for research and practice in this area.

(2 min) In this paper, we propose a novel framework to automatically utilize task-dependent semantic information which is encoded in heterogeneous information networks (HINs). Specifically, we search for a meta graph, which can capture more complex semantic relations than a meta path, to determine how graph neural networks (GNNs) propagate messages along different types of edges. We formalize the problem within the framework of neural architecture search (NAS) and then perform the search in a differentiable manner. We design an expressive search space in the form of a directed acyclic graph (DAG) to represent candidate meta graphs for a HIN, and we propose task-dependent type constraint to filter out those edge types along which message passing has no effect on the representations of nodes that are related to the downstream task. The size of the search space we define is huge, so we further propose a novel and efficient search algorithm to make the total search cost on a par with training a single GNN once. Compared with existing popular NAS algorithms, our proposed search algorithm improves the search efficiency. We conduct extensive experiments on different HINs and downstream tasks to evaluate our method, and experimental results show that our method can outperform state-of-the-art heterogeneous GNNs and also improves efficiency compared with those methods which can implicitly learn meta paths.

(2 min) The interpretation of feature importance in machine learning models is challenging when features are dependent. Permutation feature importance (PFI) ignores such dependencies, which can cause misleading interpretations due to extrapolation. A possible remedy is more advanced conditional PFI approaches that enable the assessment of feature importance conditional on all other features. Due to this shift in perspective and in order to enable correct interpretations, it is therefore important that the conditioning is transparent and humanly comprehensible. In this paper, we propose a new sampling mechanism for the conditional distribution based on permutations in conditional subgroups. As these subgroups are constructed using decision trees (transformation trees), the conditioning becomes inherently interpretable. This not only provides a simple and effective estimator of conditional PFI, but also local PFI estimates within the subgroups. In addition, we apply the conditional subgroups approach to partial dependence plots (PDP), a popular method for describing feature effects that can also suffer from extrapolation when features are dependent and interactions are present in the model. We show that PFI and PDP based on conditional subgroups often outperform methods such as conditional PFI based on knockoffs, or accumulated local effect plots. Furthermore, our approach allows for a more fine-grained interpretation of feature effects and importance within the conditional subgroups.

(2 min) Aquatic organisms can use hydrodynamic cues to navigate, find their preys and escape from predators. We consider a model of two competing microswimmers engaged in a pursue-evasion task while immersed in a low-Reynolds-number environment. The players have limited abilities: they can only sense hydrodynamic disturbances, which provide some cue about the opponent's position, and perform simple manoeuvres. The goal of the pursuer is to capturethe evader in the shortest possible time. Conversely the evader aims at deferring capture as much as possible. We show that by means of Reinforcement Learning the players find efficient and physically explainable strategies which non-trivially exploit the hydrodynamic environment. This Letter offers a proof-of-concept for the use of Reinforcement Learning to discover prey-predator strategies in aquatic environments, with potential applications to underwater robotics.

(2 min) The design of methods for inference from time sequences has traditionally relied on statistical models that describe the relation between a latent desired sequence and the observed one. A broad family of model-based algorithms have been derived to carry out inference at controllable complexity using recursive computations over the factor graph representing the underlying distribution. An alternative model-agnostic approach utilizes machine learning (ML) methods. Here we propose a framework that combines model-based algorithms and data-driven ML tools for stationary time sequences. In the proposed approach, neural networks are developed to separately learn specific components of a factor graph describing the distribution of the time sequence, rather than the complete inference task. By exploiting stationary properties of this distribution, the resulting approach can be applied to sequences of varying temporal duration. Learned factor graph can be realized using compact neural networks that are trainable using small training sets, or alternatively, be used to improve upon existing deep inference systems. We present an inference algorithm based on learned stationary factor graphs, which learns to implement the sum-product scheme from labeled data, and can be applied to sequences of different lengths. Our experimental results demonstrate the ability of the proposed learned factor graphs to learn to carry out accurate inference from small training sets for sleep stage detection using the Sleep-EDF dataset, as well as for symbol detection in digital communications with unknown channels.

(2 min) Abstraction is an important aspect of intelligence which enables agents to construct robust representations for effective decision making. In the last decade, deep networks are proven to be effective due to their ability to form increasingly complex abstractions. However, these abstractions are distributed over many neurons, making the re-use of a learned skill costly. Previous work either enforced formation of abstractions creating a designer bias, or used a large number of neural units without investigating how to obtain high-level features that may more effectively capture the source task. For avoiding designer bias and unsparing resource use, we propose to exploit neural response dynamics to form compact representations to use in skill transfer. For this, we consider two competing methods based on (1) maximum information compression principle and (2) the notion that abstract events tend to generate slowly changing signals, and apply them to the neural signals generated during task execution. To be concrete, in our simulation experiments, we either apply principal component analysis (PCA) or slow feature analysis (SFA) on the signals collected from the last hidden layer of a deep network while it performs a source task, and use these features for skill transfer in a new target task. We compare the generalization performance of these alternatives with the baselines of skill transfer with full layer output and no-transfer settings. Our results show that SFA units are the most successful for skill transfer. SFA as well as PCA, incur less resources compared to usual skill transfer, whereby many units formed show a localized response reflecting end-effector-obstacle-goal relations. Finally, SFA units with lowest eigenvalues resembles symbolic representations that highly correlate with high-level features such as joint angles which might be thought of precursors for fully symbolic systems.

(2 min) We apply methods of topological data analysis to loss functions to gain insights on learning of deep neural networks and their generalization properties. We study global properties of the loss function gradient flow. We use topological data analysis of the loss function and its Morse complex to relate local behavior along gradient trajectories with global properties of the loss surface. We define neural network Topological Obstructions score, TO-score, with help of robust topological invariants, barcodes of loss function, that quantify the badness of local minima for gradient-based optimization. We have made several experiments for computing these invariants, for small neural networks, and for fully connected, convolutional and ResNet-like neural networks on different datasets: MNIST, Fashion MNIST, CIFAR10, SVHN. Our two principal observations are as follows. Firstly, the neural network barcode and TO-score decrease with the increase of the neural network depth and width. Secondly, there is an intriguing connection between the length of minima segments in the barcode and the minima generalization error.

(2 min) Mechanistic models with differential equations are a key component of scientific applications of machine learning. Inference in such models is usually computationally demanding, because it involves repeatedly solving the differential equation. The main problem here is that the numerical solver is hard to combine with standard inference techniques. Recent work in probabilistic numerics has developed a new class of solvers for ordinary differential equations (ODEs) that phrase the solution process directly in terms of Bayesian filtering. We here show that this allows such methods to be combined very directly, with conceptual and numerical ease, with latent force models in the ODE itself. It then becomes possible to perform approximate Bayesian inference on the latent force as well as the ODE solution in a single, linear complexity pass of an extended Kalman filter / smoother - that is, at the cost of computing a single ODE solution. We demonstrate the expressiveness and performance of the algorithm by training, among others, a non-parametric SIRD model on data from the COVID-19 outbreak.

(2 min) In some industrial applications such as fraud detection, the performance of common supervision techniques may be affected by the poor quality of the available labels : in actual operational use-cases, these labels may be weak in quantity, quality or trustworthiness. We propose a benchmark to evaluate the natural robustness of different algorithms taken from various paradigms on artificially corrupted datasets, with a focus on noisy labels. This paper studies the intrinsic robustness of some leading classifiers. The algorithms under scrutiny include SVM, logistic regression, random forests, XGBoost, Khiops. Furthermore, building on results from recent literature, the study is supplemented with an investigation into the opportunity to enhance some algorithms with symmetric loss functions.

(2 min) We develop a learning-based control algorithm for unknown dynamical systems under very severe data limitations. Specifically, the algorithm has access to streaming data only from a single and ongoing trial. Despite the scarcity of data, we show -- through a series of examples -- that the algorithm can provide performance comparable to reinforcement learning algorithms trained over millions of environment interactions. It accomplishes such performance by effectively leveraging various forms of side information on the dynamics to reduce the sample complexity. Such side information typically comes from elementary laws of physics and qualitative properties of the system. More precisely, the algorithm approximately solves an optimal control problem encoding the system's desired behavior. To this end, it constructs and refines a differential inclusion that contains the unknown vector field of the dynamics. The differential inclusion, used in an interval Taylor-based method, enables to over-approximate the set of states the system may reach. Theoretically, we establish a bound on the suboptimality of the approximate solution with respect to the case of known dynamics. We show that the longer the trial or the more side information is available, the tighter the bound. Empirically, experiments in a high-fidelity F-16 aircraft simulator and MuJoCo's environments such as the Reacher, Swimmer, and Cheetah illustrate the algorithm's effectiveness.

(2 min) Meta-learning optimizes the hyperparameters of a training procedure, such as its initialization, kernel, or learning rate, based on data sampled from a number of auxiliary tasks. A key underlying assumption is that the auxiliary tasks, known as meta-training tasks, share the same generating distribution as the tasks to be encountered at deployment time, known as meta-test tasks. This may, however, not be the case when the test environment differ from the meta-training conditions. To address shifts in task generating distribution between meta-training and meta-testing phases, this paper introduces weighted free energy minimization (WFEM) for transfer meta-learning. We instantiate the proposed approach for non-parametric Bayesian regression and classification via Gaussian Processes (GPs). The method is validated on a toy sinusoidal regression problem, as well as on classification using miniImagenet and CUB data sets, through comparison with standard meta-learning of GP priors as implemented by PACOH.

(2 min) Outlier detection is a significant area in data mining. It can be either used to pre-process the data prior to an analysis or post the processing phase (before visualization) depending on the effectiveness of the outlier and its importance. Outlier detection extends to several fields such as detection of credit card fraud, network intrusions, machine failure prediction, potential terrorist attacks, and so on. Outliers are those data points with characteristics considerably different. They deviate from the data set causing inconsistencies, noise and anomalies during analysis and result in modification of the original points However, a common misconception is that outliers have to be immediately eliminated or replaced from the data set. Such points could be considered useful if analyzed separately as they could be obtained from a separate mechanism entirely making it important to the research question. This study surveys the different methods of outlier detection for spatial analysis. Spatial data or geospatial data are those that exhibit geographic properties or attributes such as position or areas. An example would be weather data such as precipitation, temperature, wind velocity, and so on collected for a defined region.

(2 min) Visual information extraction (VIE) has attracted increasing attention in recent years. The existing methods usually first organized optical character recognition (OCR) results into plain texts and then utilized token-level entity annotations as supervision to train a sequence tagging model. However, it expends great annotation costs and may be exposed to label confusion, and the OCR errors will also significantly affect the final performance. In this paper, we propose a unified weakly-supervised learning framework called TCPN (Tag, Copy or Predict Network), which introduces 1) an efficient encoder to simultaneously model the semantic and layout information in 2D OCR results; 2) a weakly-supervised training strategy that utilizes only key information sequences as supervision; and 3) a flexible and switchable decoder which contains two inference modes: one (Copy or Predict Mode) is to output key information sequences of different categories by copying a token from the input or predicting one in each time step, and the other (Tag Mode) is to directly tag the input sequence in a single forward pass. Our method shows new state-of-the-art performance on several public benchmarks, which fully proves its effectiveness.

(2 min) In this paper, we present a novel adversarial lossy video compression model. At extremely low bit-rates, standard video coding schemes suffer from unpleasant reconstruction artifacts such as blocking, ringing etc. Existing learned neural approaches to video compression have achieved reasonable success on reducing the bit-rate for efficient transmission and reduce the impact of artifacts to an extent. However, they still tend to produce blurred results under extreme compression. In this paper, we present a deep adversarial learned video compression model that minimizes an auxiliary adversarial distortion objective. We find this adversarial objective to correlate better with human perceptual quality judgement relative to traditional quality metrics such as MS-SSIM and PSNR. Our experiments using a state-of-the-art learned video compression system demonstrate a reduction of perceptual artifacts and reconstruction of detail lost especially under extremely high compression.

(2 min) Reinforcement learning is a general technique that allows an agent to learn an optimal policy and interact with an environment in sequential decision making problems. The goodness of a policy is measured by its value function starting from some initial state. The focus of this paper is to construct confidence intervals (CIs) for a policy's value in infinite horizon settings where the number of decision points diverges to infinity. We propose to model the action-value state function (Q-function) associated with a policy based on series/sieve method to derive its confidence interval. When the target policy depends on the observed data as well, we propose a SequentiAl Value Evaluation (SAVE) method to recursively update the estimated policy and its value estimator. As long as either the number of trajectories or the number of decision points diverges to infinity, we show that the proposed CI achieves nominal coverage even in cases where the optimal policy is not unique. Simulation studies are conducted to back up our theoretical findings. We apply the proposed method to a dataset from mobile health studies and find that reinforcement learning algorithms could help improve patient's health status. A Python implementation of the proposed procedure is available at https://github.com/shengzhang37/SAVE.

(2 min) To use neural networks in safety-critical settings it is paramount to provide assurances on their runtime operation. Recent work on ReLU networks has sought to verify whether inputs belonging to a bounded box can ever yield some undesirable output. Input-splitting procedures, a particular type of verification mechanism, do so by recursively partitioning the input set into smaller sets. The efficiency of these methods is largely determined by the number of splits the box must undergo before the property can be verified. In this work, we propose a new technique based on shadow prices that fully exploits the information of the problem yielding a more efficient generation of splits than the state-of-the-art. Results on the Airborne Collision Avoidance System (ACAS) benchmark verification tasks show a considerable reduction in the partitions generated which substantially reduces computation times. These results open the door to improved verification methods for a wide variety of machine learning applications including vision and control.

(3 min) It is crucial for policymakers to understand the community prevalence of COVID-19 so combative resources can be effectively allocated and prioritized during the COVID-19 pandemic. Traditionally, community prevalence has been assessed through diagnostic and antibody testing data. However, despite the increasing availability of COVID-19 testing, the required level has not been met in most parts of the globe, introducing a need for an alternative method for communities to determine disease prevalence. This is further complicated by the observation that COVID-19 prevalence and spread varies across different spatial, temporal, and demographics. In this study, we understand trends in the spread of COVID-19 by utilizing the results of self-reported COVID-19 symptoms surveys as an alternative to COVID-19 testing reports. This allows us to assess community disease prevalence, even in areas with low COVID-19 testing ability. Using individually reported symptom data from various populations, our method predicts the likely percentage of the population that tested positive for COVID-19. We do so with a Mean Absolute Error (MAE) of 1.14 and Mean Relative Error (MRE) of 60.40\% with 95\% confidence interval as (60.12, 60.67). This implies that our model predicts +/- 1140 cases than the original in a population of 1 million. In addition, we forecast the location-wise percentage of the population testing positive for the next 30 days using self-reported symptoms data from previous days. The MAE for this method is as low as 0.15 (MRE of 23.61\% with 95\% confidence interval as (23.6, 13.7)) for New York. We present an analysis of these results, exposing various clinical attributes of interest across different demographics. Lastly, we qualitatively analyze how various policy enactments (testing, curfew) affect the prevalence of COVID-19 in a community.

(2 min) We show that Transformer encoder architectures can be massively sped up, with limited accuracy costs, by replacing the self-attention sublayers with simple linear transformations that "mix" input tokens. These linear transformations, along with standard nonlinearities in feed-forward layers, prove competent at modeling semantic relationships in several text classification tasks. Most surprisingly, we find that replacing the self-attention sublayer in a Transformer encoder with a standard, unparameterized Fourier Transform achieves 92-97% of the accuracy of BERT counterparts on the GLUE benchmark, but trains nearly seven times faster on GPUs and twice as fast on TPUs. The resulting model, FNet, also scales very efficiently to long inputs. Specifically, when compared to the "efficient" Transformers on the Long Range Arena benchmark, FNet matches the accuracy of the most accurate models, but is faster than the fastest models across all sequence lengths on GPUs (and across relatively shorter lengths on TPUs). Finally, FNet has a light memory footprint and is particularly efficient at smaller model sizes: for a fixed speed and accuracy budget, small FNet models outperform Transformer counterparts.

(2 min) This letter introduces a new denoiser that modifies the structure of denoising autoencoder (DAE), namely noise learning based DAE (nlDAE). The proposed nlDAE learns the noise of the input data. Then, the denoising is performed by subtracting the regenerated noise from the noisy input. Hence, nlDAE is more effective than DAE when the noise is simpler to regenerate than the original data. To validate the performance of nlDAE, we provide three case studies: signal restoration, symbol demodulation, and precise localization. Numerical results suggest that nlDAE requires smaller latent space dimension and smaller training dataset compared to DAE.

(2 min) In this paper, we address Novel Class Discovery (NCD), the task of unveiling new classes in a set of unlabeled samples given a labeled dataset with known classes. We exploit the peculiarities of NCD to build a new framework, named Neighborhood Contrastive Learning (NCL), to learn discriminative representations that are important to clustering performance. Our contribution is twofold. First, we find that a feature extractor trained on the labeled set generates representations in which a generic query sample and its neighbors are likely to share the same class. We exploit this observation to retrieve and aggregate pseudo-positive pairs with contrastive learning, thus encouraging the model to learn more discriminative representations. Second, we notice that most of the instances are easily discriminated by the network, contributing less to the contrastive loss. To overcome this issue, we propose to generate hard negatives by mixing labeled and unlabeled samples in the feature space. We experimentally demonstrate that these two ingredients significantly contribute to clustering performance and lead our model to outperform state-of-the-art methods by a large margin (e.g., clustering accuracy +13% on CIFAR-100 and +8% on ImageNet).

(2 min) We extend the drifting games analysis to continuous time and show that the optimal adversary, if the value function has strictly positive derivative up to fourth order is bronian motion.

(2 min) EEG technology finds applications in several domains. Currently, most EEG systems require subjects to wear several electrodes on the scalp to be effective. However, several channels might include noisy information, redundant signals, induce longer preparation times and increase computational times of any automated system for EEG decoding. One way to reduce the signal-to-noise ratio and improve classification accuracy is to combine channel selection with feature extraction, but EEG signals are known to present high inter-subject variability. In this work we introduce a novel algorithm for subject-independent channel selection of EEG recordings. Considering multi-channel trial recordings as statistical units and the EEG decoding task as the class of reference, the algorithm (i) exploits channel-specific 1D-Convolutional Neural Networks (1D-CNNs) as feature extractors in a supervised fashion to maximize class separability; (ii) it reduces a high dimensional multi-channel trial representation into a unique trial vector by concatenating the channels' embeddings and (iii) recovers the complex inter-channel relationships during channel selection, by exploiting an ensemble of AutoEncoders (AE) to identify from these vectors the most relevant channels to perform classification. After training, the algorithm can be exploited by transferring only the parametrized subgroup of selected channel-specific 1D-CNNs to new signals from new subjects and obtain low-dimensional and highly informative trial vectors to be fed to any classifier.

(2 min) Popular approaches for minimizing loss in data-driven learning often involve an abstraction or an explicit retention of the history of gradients for efficient parameter updates. The aggregated history of gradients nudges the parameter updates in the right direction even when the gradients at any given step are not informative. Although the history of gradients summarized in meta-parameters or explicitly stored in memory has been shown effective in theory and practice, the question of whether $all$ or only a subset of the gradients in the history are sufficient in deciding the parameter updates remains unanswered. In this paper, we propose a framework of memory-augmented gradient descent optimizers that retain a limited view of their gradient history in their internal memory. Such optimizers scale well to large real-life datasets, and our experiments show that the memory augmented extensions of standard optimizers enjoy accelerated convergence and improved performance on a majority of computer vision and language tasks that we considered. Additionally, we prove that the proposed class of optimizers with fixed-size memory converge under assumptions of strong convexity, regardless of which gradients are selected or how they are linearly combined to form the update step.

(2 min) Graph neural networks (GNNs) model nonlinear representations in graph data with applications in distributed agent coordination, control, and planning among others. Current GNN architectures assume ideal scenarios and ignore link fluctuations that occur due to environment, human factors, or external attacks. In these situations, the GNN fails to address its distributed task if the topological randomness is not considered accordingly. To overcome this issue, we put forth the stochastic graph neural network (SGNN) model: a GNN where the distributed graph convolution module accounts for the random network changes. Since stochasticity brings in a new learning paradigm, we conduct a statistical analysis on the SGNN output variance to identify conditions the learned filters should satisfy for achieving robust transference to perturbed scenarios, ultimately revealing the explicit impact of random link losses. We further develop a stochastic gradient descent (SGD) based learning process for the SGNN and derive conditions on the learning rate under which this learning process converges to a stationary point. Numerical results corroborate our theoretical findings and compare the benefits of SGNN robust transference with a conventional GNN that ignores graph perturbations during learning.

(2 min) Cross-entropy loss with softmax output is a standard choice to train neural network classifiers. We give a new view of neural network classifiers with softmax and cross-entropy as mutual information evaluators. We show that when the dataset is balanced, training a neural network with cross-entropy maximises the mutual information between inputs and labels through a variational form of mutual information. Thereby, we develop a new form of softmax that also converts a classifier to a mutual information evaluator when the dataset is imbalanced. Experimental results show that the new form leads to better classification accuracy, in particular for imbalanced datasets.

(2 min) In supervised learning, it is known that overparameterized neural networks with one hidden layer provably and efficiently learn and generalize, when trained using stochastic gradient descent with sufficiently small learning rate and suitable initialization. In contrast, the benefit of overparameterization in unsupervised learning is not well understood. Normalizing flows (NFs) constitute an important class of models in unsupervised learning for sampling and density estimation. In this paper, we theoretically and empirically analyze these models when the underlying neural network is one-hidden-layer overparameterized network. Our main contributions are two-fold: (1) On the one hand, we provide theoretical and empirical evidence that for a class of NFs containing most of the existing NF models, overparametrization hurts training. (2) On the other hand, we prove that unconstrained NFs, a recently introduced model, can efficiently learn any reasonable data distribution under minimal assumptions when the underlying network is overparametrized.

(2 min) Conventional gradient descent methods compute the gradients for multiple variables through the partial derivative. Treating the coupled variables independently while ignoring the interaction, however, leads to an insufficient optimization for bilinear models. In this paper, we propose a dependable learning based on Cogradient Descent (CoGD) algorithm to address the bilinear optimization problem, providing a systematic way to coordinate the gradients of coupling variables based on a kernelized projection function. CoGD is introduced to solve bilinear problems when one variable is with sparsity constraint, as often occurs in modern learning paradigms. CoGD can also be used to decompose the association of features and weights, which further generalizes our method to better train convolutional neural networks (CNNs) and improve the model capacity. CoGD is applied in representative bilinear problems, including image reconstruction, image inpainting, network pruning and CNN training. Extensive experiments show that CoGD improves the state-of-the-arts by significant margins. Code is available at {https://github.com/bczhangbczhang/CoGD}.

(2 min) Even if path planning can be solved using standard techniques from dynamic programming and control, the problem can also be approached using probabilistic inference. The algorithms that emerge using the latter framework bear some appealing characteristics that qualify the probabilistic approach as a powerful alternative to the more traditional control formulations. The idea of using estimation on stochastic models to solve control problems is not new and the inference approach considered here falls under the rubric of Active Inference (AI) and Control as Inference (CAI). In this work, we look at the specific recursions that arise from various cost functions that, although they may appear similar in scope, bear noticeable differences, at least when applied to typical path planning problems. We start by posing the path planning problem on a probabilistic factor graph, and show how the various algorithms translate into specific message composition rules. We then show how this unified approach, presented both in probability space and in log space, provides a very general framework that includes the Sum-product, the Max-product, Dynamic programming and mixed Reward/Entropy criteria-based algorithms. The framework also expands algorithmic design options for smoother or sharper policy distributions, including generalized Sum/Max-product algorithm, a Smooth Dynamic programming algorithm and modified versions of the Reward/Entropy recursions. We provide a comprehensive table of recursions and a comparison through simulations, first on a synthetic small grid with a single goal with obstacles, and then on a grid extrapolated from a real-world scene with multiple goals and a semantic map.

(2 min) Network embedding has been widely used in social recommendation and network analysis, such as recommendation systems and anomaly detection with graphs. However, most of previous approaches cannot handle large graphs efficiently, due to that (i) computation on graphs is often costly and (ii) the size of graph or the intermediate results of vectors could be prohibitively large, rendering it difficult to be processed on a single machine. In this paper, we propose an efficient and effective distributed algorithm for network embedding on large graphs using Apache Spark, which recursively partitions a graph into several small-sized subgraphs to capture the internal and external structural information of nodes, and then computes the network embedding for each subgraph in parallel. Finally, by aggregating the outputs on all subgraphs, we obtain the embeddings of nodes in a linear cost. After that, we demonstrate in various experiments that our proposed approach is able to handle graphs with billions of edges within a few hours and is at least 4 times faster than the state-of-the-art approaches. Besides, it achieves up to $4.25\%$ and $4.27\%$ improvements on link prediction and node classification tasks respectively. In the end, we deploy the proposed algorithms in two online games of Tencent with the applications of friend recommendation and item recommendation, which improve the competitors by up to $91.11\%$ in running time and up to $12.80\%$ in the corresponding evaluation metrics.

(2 min) Predicting multiple heterogeneous biological and medical targets is a challenge for traditional deep learning models. In contrast to single-task learning, in which a separate model is trained for each target, multi-task learning (MTL) optimizes a single model to predict multiple related targets simultaneously. To address this challenge, we propose the Multi-gate Mixture-of-Experts with Exclusivity (MMoEEx). Our work aims to tackle the heterogeneous MTL setting, in which the same model optimizes multiple tasks with different characteristics. Such a scenario can overwhelm current MTL approaches due to the challenges in balancing shared and task-specific representations and the need to optimize tasks with competing optimization paths. Our method makes two key contributions: first, we introduce an approach to induce more diversity among experts, thus creating representations more suitable for highly imbalanced and heterogenous MTL learning; second, we adopt a two-step optimization [6, 11] approach to balancing the tasks at the gradient level. We validate our method on three MTL benchmark datasets, including Medical Information Mart for Intensive Care (MIMIC-III) and PubChem BioAssay (PCBA).

(2 min) Compositional generalization is the ability to generalize systematically to a new data distribution by combining known components. Although humans seem to have a great ability to generalize compositionally, state-of-the-art neural models struggle to do so. In this work, we study compositional generalization in classification tasks and present two main contributions. First, we study ways to convert a natural language sequence-to-sequence dataset to a classification dataset that also requires compositional generalization. Second, we show that providing structural hints (specifically, providing parse trees and entity links as attention masks for a Transformer model) helps compositional generalization.

(2 min) Automatic generation of high-quality commit messages for code commits can substantially facilitate software developers' works and coordination. However, the semantic gap between source code and natural language poses a major challenge for the task. Several studies have been proposed to alleviate the challenge but none explicitly involves code contextual information during commit message generation. Specifically, existing research adopts static embedding for code tokens, which maps a token to the same vector regardless of its context. In this paper, we propose a novel Contextualized code representation learning strategy for commit message Generation (CoreGen). CoreGen first learns contextualized code representations which exploit the contextual information behind code commit sequences. The learned representations of code commits built upon Transformer are then fine-tuned for downstream commit message generation. Experiments on the benchmark dataset demonstrate the superior effectiveness of our model over the baseline models with at least 28.18% improvement in terms of BLEU-4 score. Furthermore, we also highlight the future opportunities in training contextualized code representations on larger code corpus as a solution to low-resource tasks and adapting the contextualized code representation framework to other code-to-text generation tasks.

(2 min) Tuning complex machine learning systems is challenging. Machine learning typically requires to set hyperparameters, be it regularization, architecture, or optimization parameters, whose tuning is critical to achieve good predictive performance. To democratize access to machine learning systems, it is essential to automate the tuning. This paper presents Amazon SageMaker Automatic Model Tuning (AMT), a fully managed system for gradient-free optimization at scale. AMT finds the best version of a trained machine learning model by repeatedly evaluating it with different hyperparameter configurations. It leverages either random search or Bayesian optimization to choose the hyperparameter values resulting in the best model, as measured by the metric chosen by the user. AMT can be used with built-in algorithms, custom algorithms, and Amazon SageMaker pre-built containers for machine learning frameworks. We discuss the core functionality, system architecture, our design principles, and lessons learned. We also describe more advanced features of AMT, such as automated early stopping and warm-starting, showing in experiments their benefits to users.

(2 min) Contrastive loss and its variants have become very popular recently for learning visual representations without supervision. In this work, we study three intriguing properties of contrastive learning. We first generalize the standard contrastive loss to a broader family of losses, and we find that various instantiations of the generalized loss perform similarly under the presence of a multi-layer non-linear projection head. We then study if instance-based contrastive learning (such as in SimCLR, MoCo, BYOL, and so on, which are based on global image representation) can learn well on images with multiple objects present. We find that meaningful hierarchical local features can be learned despite the fact that these objectives operate on global instance-level features. Finally, we study an intriguing phenomenon of feature suppression among competing features shared across augmented views, such as "color distribution" vs "object class". We construct datasets with explicit and controllable competing features, and show that, for contrastive learning, a few bits of easy-to-learn shared features can suppress, and even fully prevent, the learning of other sets of competing features. In scenarios where there are multiple objects in an image, the dominant object would suppress the learning of smaller objects. Existing contrastive learning methods critically rely on data augmentation to favor certain sets of features over others, and face potential limitation for scenarios where existing augmentations cannot fully address the feature suppression. This poses open challenges to existing contrastive learning techniques.

(2 min) Data programming (DP) has proven to be an attractive alternative to costly hand-labeling of data. In DP, users encode domain knowledge into \emph{labeling functions} (LF), heuristics that label a subset of the data noisily and may have complex dependencies. A label model is then fit to the LFs to produce an estimate of the unknown class label. The effects of label model misspecification on test set performance of a downstream classifier are understudied. This presents a serious awareness gap to practitioners, in particular since the dependency structure among LFs is frequently ignored in field applications of DP. We analyse modeling errors due to structure over-specification. We derive novel theoretical bounds on the modeling error and empirically show that this error can be substantial, even when modeling a seemingly sensible structure.

(2 min) Meta-learning (ML) has emerged as a promising direction in learning models under constrained resource settings like few-shot learning. The popular approaches for ML either learn a generalizable initial model or a generic parametric optimizer through episodic training. The former approaches leverage the knowledge from a batch of tasks to learn an optimal prior. In this work, we study the importance of a batch for ML. Specifically, we first incorporate a batch episodic training regimen to improve the learning of the generic parametric optimizer. We also hypothesize that the common assumption in batch episodic training that each task in a batch has an equal contribution to learning an optimal meta-model need not be true. We propose to weight the tasks in a batch according to their "importance" in improving the meta-model's learning. To this end, we introduce a training curriculum motivated by selective focus in humans, called task attended meta-training, to weight the tasks in a batch. Task attention is a standalone module that can be integrated with any batch episodic training regimen. The comparisons of the models with their non-task-attended counterparts on complex datasets like miniImageNet and tieredImageNet validate its effectiveness.

(2 min) This paper is dedicated to designing provably efficient adversarial imitation learning (AIL) algorithms that directly optimize policies from expert demonstrations. Firstly, we develop a transition-aware AIL algorithm named TAIL with an expert sample complexity of $\tilde{O}(H^{3/2} |S|/\varepsilon)$ under the known transition setting, where $H$ is the planning horizon, $|S|$ is the state space size and $\varepsilon$ is desired policy value gap. This improves upon the previous best bound of $\tilde{O}(H^2 |S| / \varepsilon^2)$ for AIL methods and matches the lower bound of $\tilde{\Omega} (H^{3/2} |S|/\varepsilon)$ in [Rajaraman et al., 2021] up to a logarithmic factor. The key ingredient of TAIL is a fine-grained estimator for expert state-action distribution, which explicitly utilizes the transition function information. Secondly, considering practical settings where the transition functions are usually unknown but environment interaction is allowed, we accordingly develop a model-based transition-aware AIL algorithm named MB-TAIL. In particular, MB-TAIL builds an empirical transition model by interacting with the environment and performs imitation under the recovered empirical model. The interaction complexity of MB-TAIL is $\tilde{O} (H^3 |S|^2 |A| / \varepsilon^2)$, which improves the best known result of $\tilde{O} (H^4 |S|^2 |A| / \varepsilon^2)$ in [Shani et al., 2021]. Finally, our theoretical results are supported by numerical evaluation and detailed analysis on two challenging MDPs.

(2 min) Graph convolutional neural networks (GCNNs) are nonlinear processing tools to learn representations from network data. A key property of GCNNs is their stability to graph perturbations. Current analysis considers deterministic perturbations but fails to provide relevant insights when topological changes are random. This paper investigates the stability of GCNNs to stochastic graph perturbations induced by link losses. In particular, it proves the expected output difference between the GCNN over random perturbed graphs and the GCNN over the nominal graph is upper bounded by a factor that is linear in the link loss probability. We perform the stability analysis in the graph spectral domain such that the result holds uniformly for any graph. This result also shows the role of the nonlinearity and the architecture width and depth, and allows identifying handle to improve the GCNN robustness. Numerical simulations on source localization and robot swarm control corroborate our theoretical findings.

(2 min) We employ constraints to control the parameter space of deep neural networks throughout training. The use of customized, appropriately designed constraints can reduce the vanishing/exploding gradients problem, improve smoothness of classification boundaries, control weight magnitudes and stabilize deep neural networks, and thus enhance the robustness of training algorithms and the generalization capabilities of neural networks. We provide a general approach to efficiently incorporate constraints into a stochastic gradient Langevin framework, allowing enhanced exploration of the loss landscape. We also present specific examples of constrained training methods motivated by orthogonality preservation for weight matrices and explicit weight normalizations. Discretization schemes are provided both for the overdamped formulation of Langevin dynamics and the underdamped form, in which momenta further improve sampling efficiency. These optimization schemes can be used directly, without needing to adapt neural network architecture design choices or to modify the objective with regularization terms, and see performance improvements in classification tasks.

(2 min) Dynamic transportation networks have been analyzed for years by means of static graph-based indicators in order to study the temporal evolution of relevant network components, and to reveal complex dependencies that would not be easily detected by a direct inspection of the data. This paper presents a state-of-the-art latent network model to forecast multilayer dynamic graphs that are increasingly common in transportation and proposes a community-based extension to reduce the computational burden. Flexible time series analysis is obtained by modeling the probability of edges between vertices through latent Gaussian processes. The models and Bayesian inference are illustrated on a sample of 10-year data from four major airlines within the US air transportation system. Results show how the estimated latent parameters from the models are related to the airline's connectivity dynamics, and their ability to project the multilayer graph into the future for out-of-sample full network forecasts, while stochastic blockmodeling allows for the identification of relevant communities. Reliable network predictions would allow policy-makers to better understand the dynamics of the transport system, and help in their planning on e.g. route development, or the deployment of new regulations.

(2 min) In this paper, we investigate the non-stationary combinatorial semi-bandit problem, both in the switching case and in the dynamic case. In the general case where (a) the reward function is non-linear, (b) arms may be probabilistically triggered, and (c) only approximate offline oracle exists \cite{wang2017improving}, our algorithm achieves $\tilde{\mathcal{O}}(\sqrt{\mathcal{S} T})$ distribution-dependent regret in the switching case, and $\tilde{\mathcal{O}}(\mathcal{V}^{1/3}T^{2/3})$ in the dynamic case, where $\mathcal S$ is the number of switchings and $\mathcal V$ is the sum of the total ``distribution changes''. The regret bounds in both scenarios are nearly optimal, but our algorithm needs to know the parameter $\mathcal S$ or $\mathcal V$ in advance. We further show that by employing another technique, our algorithm no longer needs to know the parameters $\mathcal S$ or $\mathcal V$ but the regret bounds could become suboptimal. In a special case where the reward function is linear and we have an exact oracle, we design a parameter-free algorithm that achieves nearly optimal regret both in the switching case and in the dynamic case without knowing the parameters in advance.

(2 min) Differential privacy is a restriction on data processing algorithms that provides strong confidentiality guarantees for individual records in the data. However, research on proper statistical inference, that is, research on properly quantifying the uncertainty of the (noisy) sample estimate regarding the true value in the population, is currently still limited. This paper proposes and evaluates several strategies to compute valid differentially private confidence intervals for the median. Instead of computing a differentially private point estimate and deriving its uncertainty, we directly estimate the interval bounds and discuss why this approach is superior if ensuring privacy is important. We also illustrate that addressing both sources of uncertainty--the error from sampling and the error from protecting the output--simultaneously should be preferred over simpler approaches that incorporate the uncertainty in a sequential fashion. We evaluate the performance of the different algorithms under various parameter settings in extensive simulation studies and demonstrate how the findings could be applied in practical settings using data from the 1940 Decennial Census.

(2 min) Understanding the reasons associated with successful proposals is of paramount importance to improve evaluation processes. In this context, we analyzed whether bibliometric features are able to predict the success of research grants. We extracted features aiming at characterizing the academic history of Brazilian researchers, including research topics, affiliations, number of publications and visibility. The extracted features were then used to predict grants productivity via machine learning in three major research areas, namely Medicine, Dentistry and Veterinary Medicine. We found that research subject and publication history play a role in predicting productivity. In addition, institution-based features turned out to be relevant when combined with other features. While the best results outperformed text-based attributes, the evaluated features were not highly discriminative. Our findings indicate that predicting grants success, at least with the considered set of bibliometric features, is not a trivial task.

(2 min) We offer straightforward theoretical results that justify incorporating machine learning in the standard linear instrumental variable setting. The key idea is to use machine learning, combined with sample-splitting, to predict the treatment variable from the instrument and any exogenous covariates, and then use this predicted treatment and the covariates as technical instruments to recover the coefficients in the second-stage. This allows the researcher to extract non-linear co-variation between the treatment and instrument that may dramatically improve estimation precision and robustness by boosting instrument strength. Importantly, we constrain the machine-learned predictions to be linear in the exogenous covariates, thus avoiding spurious identification arising from non-linear relationships between the treatment and the covariates. We show that this approach delivers consistent and asymptotically normal estimates under weak conditions and that it may be adapted to be semiparametrically efficient (Chamberlain, 1992). Our method preserves standard intuitions and interpretations of linear instrumental variable methods, including under weak identification, and provides a simple, user-friendly upgrade to the applied economics toolbox. We illustrate our method with an example in law and criminal justice, examining the causal effect of appellate court reversals on district court sentencing decisions.

(2 min) We show a simple reduction which demonstrates the cryptographic hardness of learning a single periodic neuron over isotropic Gaussian distributions in the presence of noise. More precisely, our reduction shows that any polynomial-time algorithm (not necessarily gradient-based) for learning such functions under small noise implies a polynomial-time quantum algorithm for solving worst-case lattice problems, whose hardness form the foundation of lattice-based cryptography. Our core hard family of functions, which are well-approximated by one-layer neural networks, take the general form of a univariate periodic function applied to an affine projection of the data. These functions have appeared in previous seminal works which demonstrate their hardness against gradient-based (Shamir'18), and Statistical Query (SQ) algorithms (Song et al.'17). We show that if (polynomially) small noise is added to the labels, the intractability of learning these functions applies to all polynomial-time algorithms under the aforementioned cryptographic assumptions. Moreover, we demonstrate the necessity of noise in the hardness result by designing a polynomial-time algorithm for learning certain families of such functions under exponentially small adversarial noise. Our proposed algorithm is not a gradient-based or an SQ algorithm, but is rather based on the celebrated Lenstra-Lenstra-Lov\'asz (LLL) lattice basis reduction algorithm. Furthermore, in the absence of noise, this algorithm can be directly applied to solve CLWE detection (Bruna et al.'21) and phase retrieval with an optimal sample complexity of $d+1$ samples. In the former case, this improves upon the quadratic-in-$d$ sample complexity required in (Bruna et al.'21). In the latter case, this improves upon the state-of-the-art AMP-based algorithm, which requires approximately $1.128d$ samples (Barbier et al.'19).

(2 min) Shapley Value is commonly adopted to measure and incentivize client participation in federated learning. In this paper, we show -- theoretically and through simulations -- that Shapley Value underestimates the contribution of a common type of client: the Maverick. Mavericks are clients that differ both in data distribution and data quantity and can be the sole owners of certain types of data. Selecting the right clients at the right moment is important for federated learning to reduce convergence times and improve accuracy. We propose FedEMD, an adaptive client selection strategy based on the Wasserstein distance between the local and global data distributions. As FedEMD adapts the selection probability such that Mavericks are preferably selected when the model benefits from improvement on rare classes, it consistently ensures the fast convergence in the presence of different types of Mavericks. Compared to existing strategies, including Shapley Value-based ones, FedEMD improves the convergence of neural network classifiers by at least 26.9% for FedAvg aggregation compared with the state of the art.

(2 min) For the last few decades, classical machine learning has allowed us to improve the lives of many through automation, natural language processing, predictive analytics and much more. However, a major concern is the fact that we're fast approach the threshold of the maximum possible computational capacity available to us by the means of classical computing devices including CPUs, GPUs and Application Specific Integrated Circuits (ASICs). This is due to the exponential increase in model sizes which now have parameters in the magnitude of billions and trillions, requiring a significant amount of computing resources across a significant amount of time, just to converge one single model. To observe the efficacy of using quantum computing for certain machine learning tasks and explore the improved potential of convergence, error reduction and robustness to noisy data, this paper will look forth to test and verify the aspects in which quantum machine learning can help improve over classical machine learning approaches while also shedding light on the likely limitations that have prevented quantum approaches to become the mainstream. A major focus will be to recreate the work by Farhi et al and conduct experiments using their theory of performing machine learning in a quantum context, with assistance from the Tensorflow Quantum documentation.

(2 min) Robust training methods against perturbations to the input data have received great attention in the machine learning literature. A standard approach in this direction is adversarial training which learns a model using adversarially-perturbed training samples. However, adversarial training performs suboptimally against perturbations structured across samples such as universal and group-sparse shifts that are commonly present in biological data such as gene expression levels of different tissues. In this work, we seek to close this optimality gap and introduce Group-Structured Adversarial Training (GSAT) which learns a model robust to perturbations structured across samples. We formulate GSAT as a non-convex concave minimax optimization problem which minimizes a group-structured optimal transport cost. Specifically, we focus on the applications of GSAT for group-sparse and rank-constrained perturbations modeled using group and nuclear norm penalties. In order to solve GSAT's non-smooth optimization problem in those cases, we propose a new minimax optimization algorithm called GDADMM by combining Gradient Descent Ascent (GDA) and Alternating Direction Method of Multipliers (ADMM). We present several applications of the GSAT framework to gain robustness against structured perturbations for image recognition and computational biology datasets.

(2 min) Backpropagation image saliency aims at explaining model predictions by estimating model-centric importance of individual pixels in the input. However, class-insensitivity of the earlier layers in a network only allows saliency computation with low resolution activation maps of the deeper layers, resulting in compromised image saliency. Remedifying this can lead to sanity failures. We propose CAMERAS, a technique to compute high-fidelity backpropagation saliency maps without requiring any external priors and preserving the map sanity. Our method systematically performs multi-scale accumulation and fusion of the activation maps and backpropagated gradients to compute precise saliency maps. From accurate image saliency to articulation of relative importance of input features for different models, and precise discrimination between model perception of visually similar objects, our high-resolution mapping offers multiple novel insights into the black-box deep visual models, which are presented in the paper. We also demonstrate the utility of our saliency maps in adversarial setup by drastically reducing the norm of attack signals by focusing them on the precise regions identified by our maps. Our method also inspires new evaluation metrics and a sanity check for this developing research direction. Code is available here https://github.com/VisMIL/CAMERAS

(2 min) In this paper, we propose a new method to overcome catastrophic forgetting by adding generative regularization to Bayesian inference framework. Bayesian method provides a general framework for continual learning. We could further construct a generative regularization term for all given classification models by leveraging energy-based models and Langevin-dynamic sampling to enrich the features learned in each task. By combining discriminative and generative loss together, we empirically show that the proposed method outperforms state-of-the-art methods on a variety of tasks, avoiding catastrophic forgetting in continual learning. In particular, the proposed method outperforms baseline methods over 15% on the Fashion-MNIST dataset and 10% on the CUB dataset

(2 min) Solving the optimal power flow (OPF) problem in real-time electricity market improves the efficiency and reliability in the integration of low-carbon energy resources into the power grids. To address the scalability and adaptivity issues of existing end-to-end OPF learning solutions, we propose a new graph neural network (GNN) framework for predicting the electricity market prices from solving OPFs. The proposed GNN-for-OPF framework innovatively exploits the locality property of prices and introduces physics-aware regularization, while attaining reduced model complexity and fast adaptivity to varying grid topology. Numerical tests have validated the learning efficiency and adaptivity improvements of our proposed method over existing approaches.

(2 min) Works on lottery ticket hypothesis (LTH) and single-shot network pruning (SNIP) have raised a lot of attention currently on post-training pruning (iterative magnitude pruning), and before-training pruning (pruning at initialization). The former method suffers from an extremely large computation cost and the latter category of methods usually struggles with insufficient performance. In comparison, during-training pruning, a class of pruning methods that simultaneously enjoys the training/inference efficiency and the comparable performance, temporarily, has been less explored. To better understand during-training pruning, we quantitatively study the effect of pruning throughout training from the perspective of pruning plasticity (the ability of the pruned networks to recover the original performance). Pruning plasticity can help explain several other empirical observations about neural network pruning in literature. We further find that pruning plasticity can be substantially improved by injecting a brain-inspired mechanism called neuroregeneration, i.e., to regenerate the same number of connections as pruned. Based on the insights from pruning plasticity, we design a novel gradual magnitude pruning (GMP) method, named gradual pruning with zero-cost neuroregeneration (GraNet), and its dynamic sparse training (DST) variant (GraNet-ST). Both of them advance state of the art. Perhaps most impressively, the latter for the first time boosts the sparse-to-sparse training performance over various dense-to-sparse methods by a large margin with ResNet-50 on ImageNet. We will release all codes.

(2 min) In the status quo, dementia is yet to be cured. Precise diagnosis prior to the onset of the symptoms can prevent the rapid progression of the emerging cognitive impairment. Recent progress has shown that Electroencephalography (EEG) is the promising and cost-effective test to facilitate the detection of neurocognitive disorders. However, most of the existing works have been using only resting-state EEG. The efficiencies of EEG signals from various cognitive tasks, for dementia classification, have yet to be thoroughly investigated. In this study, we designed four cognitive tasks that engage different cognitive performances: attention, working memory, and executive function. We investigated these tasks by using statistical analysis on both time and frequency domains of EEG signals from three classes of human subjects: Dementia (DEM), Mild Cognitive Impairment (MCI), and Normal Control (NC). We also further evaluated the classification performances of two features extraction methods: Principal Component Analysis (PCA) and Filter Bank Common Spatial Pattern (FBCSP). We found that the working memory related tasks yielded good performances for dementia recognition in both cases using PCA and FBCSP. Moreover, FBCSP with features combination from four tasks revealed the best sensitivity of 0.87 and the specificity of 0.80. To our best knowledge, this is the first work that concurrently investigated several cognitive tasks for dementia recognition using both statistical analysis and classification scores. Our results give essential information to design and aid in conducting further experimental tasks to early diagnose dementia patients.

(2 min) Offline reinforcement learning (RL) tries to learn the near-optimal policy with recorded offline experience without online exploration. Current offline RL research includes: 1) generative modeling, i.e., approximating a policy using fixed data; and 2) learning the state-action value function. While most research focuses on the state-action function part through reducing the bootstrapping error in value function approximation induced by the distribution shift of training data, the effects of error propagation in generative modeling have been neglected. In this paper, we analyze the error in generative modeling. We propose AQL (action-conditioned Q-learning), a residual generative model to reduce policy approximation error for offline RL. We show that our method can learn more accurate policy approximations in different benchmark datasets. In addition, we show that the proposed offline RL method can learn more competitive AI agents in complex control tasks under the multiplayer online battle arena (MOBA) game Honor of Kings.

(2 min) How can we find a subset of training samples that are most responsible for a specific prediction made by a complex black-box machine learning model? More generally, how can we explain the model's decisions to end-users in a transparent way? We propose a new model-agnostic algorithm to identify a minimal set of training samples that are indispensable for a given model's decision at a particular test point, i.e., the model's decision would have changed upon the removal of this subset from the training dataset. Our algorithm identifies such a set of "indispensable" samples iteratively by solving a constrained optimization problem. Further, we speed up the algorithm through efficient approximations and provide theoretical justification for its performance. To demonstrate the applicability and effectiveness of our approach, we apply it to a variety of tasks including data poisoning detection, training set debugging and understanding loan decisions. The results show that our algorithm is an effective and easy-to-comprehend tool that helps to better understand local model behavior, and therefore facilitates the adoption of machine learning in domains where such understanding is a requisite.

(2 min) Motivated by applications in crowdsourced entity resolution in database, signed edge prediction in social networks and correlation clustering, Mazumdar and Saha [NIPS 2017] proposed an elegant theoretical model for studying clustering with a faulty oracle. In this model, given a set of $n$ items which belong to $k$ unknown groups (or clusters), our goal is to recover the clusters by asking pairwise queries to an oracle. This oracle can answer the query that ``do items $u$ and $v$ belong to the same cluster?''. However, the answer to each pairwise query errs with probability $\varepsilon$, for some $\varepsilon\in(0,\frac12)$. Mazumdar and Saha provided two algorithms under this model: one algorithm is query-optimal while time-inefficient (i.e., running in quasi-polynomial time), the other is time efficient (i.e., in polynomial time) while query-suboptimal. Larsen, Mitzenmacher and Tsourakakis [WWW 2020] then gave a new time-efficient algorithm for the special case of $2$ clusters, which is query-optimal if the bias $\delta:=1-2\varepsilon$ of the model is large. It was left as an open question whether one can obtain a query-optimal, time-efficient algorithm for the general case of $k$ clusters and other regimes of $\delta$. In this paper, we make progress on the above question and provide a time-efficient algorithm with nearly-optimal query complexity (up to a factor of $O(\log^2 n)$) for all constant $k$ and any $\delta$ in the regime when information-theoretic recovery is possible. Our algorithm is built on a connection to the stochastic block model.

(2 min) Early and reliable COVID-19 diagnosis based on chest 3-D CT scans can assist medical specialists in vital circumstances. Deep learning methodologies constitute a main approach for chest CT scan analysis and disease prediction. However, large annotated databases are necessary for developing deep learning models that are able to provide COVID-19 diagnosis across various medical environments in different countries. Due to privacy issues, publicly available COVID-19 CT datasets are highly difficult to obtain, which hinders the research and development of AI-enabled diagnosis methods of COVID-19 based on CT scans. In this paper we present the COV19-CT-DB database which is annotated for COVID-19, consisting of about 5,000 3-D CT scans, We have split the database in training, validation and test datasets. The former two datasets can be used for training and validation of machine learning models, while the latter will be used for evaluation of the developed models. We also present a deep learning approach, based on a CNN-RNN network and report its performance on the COVID19-CT-DB database.

(2 min) We empirically demonstrate that full-batch gradient descent on neural network training objectives typically operates in a regime we call the Edge of Stability. In this regime, the maximum eigenvalue of the training loss Hessian hovers just above the numerical value $2 / \text{(step size)}$, and the training loss behaves non-monotonically over short timescales, yet consistently decreases over long timescales. Since this behavior is inconsistent with several widespread presumptions in the field of optimization, our findings raise questions as to whether these presumptions are relevant to neural network training. We hope that our findings will inspire future efforts aimed at rigorously understanding optimization at the Edge of Stability. Code is available at https://github.com/locuslab/edge-of-stability.

(2 min) In the context of lossy compression, Blau & Michaeli (2019) adopt a mathematical notion of perceptual quality and define the information rate-distortion-perception function, generalizing the classical rate-distortion tradeoff. We consider the notion of universal representations in which one may fix an encoder and vary the decoder to achieve any point within a collection of distortion and perception constraints. We prove that the corresponding information-theoretic universal rate-distortion-perception function is operationally achievable in an approximate sense. Under MSE distortion, we show that the entire distortion-perception tradeoff of a Gaussian source can be achieved by a single encoder of the same rate asymptotically. We then characterize the achievable distortion-perception region for a fixed representation in the case of arbitrary distributions, identify conditions under which the aforementioned results continue to hold approximately, and study the case when the rate is not fixed in advance. This motivates the study of practical constructions that are approximately universal across the RDP tradeoff, thereby alleviating the need to design a new encoder for each objective. We provide experimental results on MNIST and SVHN suggesting that on image compression tasks, the operational tradeoffs achieved by machine learning models with a fixed encoder suffer only a small penalty when compared to their variable encoder counterparts.

(2 min) Automatic forecasting is the task of receiving a time series and returning a forecast for the next time steps without any human intervention. Gaussian Processes (GPs) are a powerful tool for modeling time series, but so far there are no competitive approaches for automatic forecasting based on GPs. We propose practical solutions to two problems: automatic selection of the optimal kernel and reliable estimation of the hyperparameters. We propose a fixed composition of kernels, which contains the components needed to model most time series: linear trend, periodic patterns, and other flexible kernel for modeling the non-linear trend. Not all components are necessary to model each time series; during training the unnecessary components are automatically made irrelevant via automatic relevance determination (ARD). We moreover assign priors to the hyperparameters, in order to keep the inference within a plausible range; we design such priors through an empirical Bayes approach. We present results on many time series of different types; our GP model is more accurate than state-of-the-art time series models. Thanks to the priors, a single restart is enough the estimate the hyperparameters; hence the model is also fast to train.

(2 min) Increasing use of sensor data in intelligent transportation systems calls for accurate imputation algorithms that can enable reliable traffic management in the occasional absence of data. As one of the effective imputation approaches, generative adversarial networks (GANs) are implicit generative models that can be used for data imputation, which is formulated as an unsupervised learning problem. This work introduces a novel iterative GAN architecture, called Iterative Generative Adversarial Networks for Imputation (IGANI), for data imputation. IGANI imputes data in two steps and maintains the invertibility of the generative imputer, which will be shown to be a sufficient condition for the convergence of the proposed GAN-based imputation. The performance of our proposed method is evaluated on (1) the imputation of traffic speed data collected in the city of Guangzhou in China, and the training of short-term traffic prediction models using imputed data, and (2) the imputation of multi-variable traffic data of highways in Portland-Vancouver metropolitan region which includes volume, occupancy, and speed with different missing rates for each of them. It is shown that our proposed algorithm mostly produces more accurate results compared to those of previous GAN-based imputation architectures.

(2 min) We develop a new method of online inference for a vector of parameters estimated by the Polyak-Ruppert averaging procedure of stochastic gradient descent (SGD) algorithms. We leverage insights from time series regression in econometrics and construct asymptotically pivotal statistics via random scaling. Our approach is fully operational with online data and is rigorously underpinned by a functional central limit theorem. Our proposed inference method has a couple of key advantages over the existing methods. First, the test statistic is computed in an online fashion with only SGD iterates and the critical values can be obtained without any resampling methods, thereby allowing for efficient implementation suitable for massive online data. Second, there is no need to estimate the asymptotic variance and our inference method is shown to be robust to changes in the tuning parameters for SGD algorithms in simulation experiments with synthetic data.

(2 min) Deep neural networks are widespread due to their powerful performance. Yet, they suffer from degraded performance in the presence of noisy labels at train time or adversarial examples during inference. Inspired by the setting of learning with expert advice, where multiplicative weights (MW) updates were recently shown to be robust to moderate adversarial corruptions, we propose to use MW for reweighting examples during neural networks optimization. We establish the convergence of our method when used with gradient descent and show its advantage in two simple examples. We then validate empirically our findings by demonstrating that MW improve networks accuracy in the presence of label noise on CIFAR-10, CIFAR-100 and Clothing1M, and leads to better robustness to adversarial attacks.

(2 min) Fact-based Visual Question Answering (FVQA), a challenging variant of VQA, requires a QA-system to include facts from a diverse knowledge graph (KG) in its reasoning process to produce an answer. Large KGs, especially common-sense KGs, are known to be incomplete, i.e., not all non-existent facts are always incorrect. Therefore, being able to reason over incomplete KGs for QA is a critical requirement in real-world applications that has not been addressed extensively in the literature. We develop a novel QA architecture that allows us to reason over incomplete KGs, something current FVQA state-of-the-art (SOTA) approaches lack due to their critical reliance on fact retrieval. We use KG Embeddings, a technique widely used for KG completion, for the downstream task of FVQA. We also employ a new image representation technique we call 'Image-as-Knowledge' to enable this capability, alongside a simple one-step CoAttention mechanism to attend to text and image during QA. Our FVQA architecture is faster during inference time, being O(m), as opposed to existing FVQA SOTA methods which are O(N log N), where m = number of vertices, N = number of edges = O(m^2). KG embeddings are shown to hold complementary information to word embeddings: a combination of both metrics permits performance comparable to SOTA methods in the standard answer retrieval task, and significantly better (26% absolute) in the proposed missing-edge reasoning task.

(2 min) Communication overhead is the key challenge for distributed training. Gradient compression is a widely used approach to reduce communication traffic. When combining with parallel communication mechanism method like pipeline, gradient compression technique can greatly alleviate the impact of communication overhead. However, there exists two problems of gradient compression technique to be solved. Firstly, gradient compression brings in extra computation cost, which will delay the next training iteration. Secondly, gradient compression usually leads to the decrease of convergence accuracy.

(2 min) Sentiment and lexical analyses are widely used to detect depression or anxiety disorders. It has been documented that there are significant differences in the language used by a person with emotional disorders in comparison to a healthy individual. Still, the effectiveness of these lexical approaches could be improved further because the current analysis focuses on what the social media entries are about, and not how they are written. In this study, we focus on aspects in which these short texts are similar to each other, and how they were created. We present an innovative approach to the depression screening problem by applying Collgram analysis, which is a known effective method of obtaining linguistic information from texts. We compare these results with sentiment analysis based on the BERT architecture. Finally, we create a hybrid model achieving a diagnostic accuracy of 71%.

(2 min) Existing tensor completion formulation mostly relies on partial observations from a single tensor. However, tensors extracted from real-world data are often more complex due to: (i) Partial observation: Only a small subset (e.g., 5%) of tensor elements are available. (ii) Coarse observation: Some tensor modes only present coarse and aggregated patterns (e.g., monthly summary instead of daily reports). In this paper, we are given a subset of the tensor and some aggregated/coarse observations (along one or more modes) and seek to recover the original fine-granular tensor with low-rank factorization. We formulate a coupled tensor completion problem and propose an efficient Multi-resolution Tensor Completion model (MTC) to solve the problem. Our MTC model explores tensor mode properties and leverages the hierarchy of resolutions to recursively initialize an optimization setup, and optimizes on the coupled system using alternating least squares. MTC ensures low computational and space complexity. We evaluate our model on two COVID-19 related spatio-temporal tensors. The experiments show that MTC could provide 65.20% and 75.79% percentage of fitness (PoF) in tensor completion with only 5% fine granular observations, which is 27.96% relative improvement over the best baseline. To evaluate the learned low-rank factors, we also design a tensor prediction task for daily and cumulative disease case predictions, where MTC achieves 50% in PoF and 30% relative improvements over the best baseline.

(2 min) Path planning, the problem of efficiently discovering high-reward trajectories, often requires optimizing a high-dimensional and multimodal reward function. Popular approaches like CEM and CMA-ES greedily focus on promising regions of the search space and may get trapped in local maxima. DOO and VOOT balance exploration and exploitation, but use space partitioning strategies independent of the reward function to be optimized. Recently, LaMCTS empirically learns to partition the search space in a reward-sensitive manner for black-box optimization. In this paper, we develop a novel formal regret analysis for when and why such an adaptive region partitioning scheme works. We also propose a new path planning method PlaLaM which improves the function value estimation within each sub-region, and uses a latent representation of the search space. Empirically, PlaLaM outperforms existing path planning methods in 2D navigation tasks, especially in the presence of difficult-to-escape local optima, and shows benefits when plugged into model-based RL with planning components such as PETS. These gains transfer to highly multimodal real-world tasks, where we outperform strong baselines in compiler phase ordering by up to 245% and in molecular design by up to 0.4 on properties on a 0-1 scale.

(2 min) Federated learning is the distributed machine learning framework that enables collaborative training across multiple parties while ensuring data privacy. Practical adaptation of XGBoost, the state-of-the-art tree boosting framework, to federated learning remains limited due to high cost incurred by conventional privacy-preserving methods. To address the problem, we propose two variants of federated XGBoost with privacy guarantee: FedXGBoost-SMM and FedXGBoost-LDP. Our first protocol FedXGBoost-SMM deploys enhanced secure matrix multiplication method to preserve privacy with lossless accuracy and lower overhead than encryption-based techniques. Developed independently, the second protocol FedXGBoost-LDP is heuristically designed with noise perturbation for local differential privacy, and empirically evaluated on real-world and synthetic datasets.

(2 min) Despite their widespread success, the application of deep neural networks to functional data remains scarce today. The infinite dimensionality of functional data means standard learning algorithms can be applied only after appropriate dimension reduction, typically achieved via basis expansions. Currently, these bases are chosen a priori without the information for the task at hand and thus may not be effective for the designated task. We instead propose to adaptively learn these bases in an end-to-end fashion. We introduce neural networks that employ a new Basis Layer whose hidden units are each basis functions themselves implemented as a micro neural network. Our architecture learns to apply parsimonious dimension reduction to functional inputs that focuses only on information relevant to the target rather than irrelevant variation in the input function. Across numerous classification/regression tasks with functional data, our method empirically outperforms other types of neural networks, and we prove that our approach is statistically consistent with low generalization error. Code is available at: \url{https://github.com/jwyyy/AdaFNN}.

(2 min) Formulating real-world optimization problems often begins with making predictions from historical data (e.g., an optimizer that aims to recommend fast routes relies upon travel-time predictions). Typically, learning the prediction model used to generate the optimization problem and solving that problem are performed in two separate stages. Recent work has showed how such prediction models can be learned end-to-end by differentiating through the optimization task. Such methods often yield empirical improvements, which are typically attributed to end-to-end making better error tradeoffs than the standard loss function used in a two-stage solution. We refine this explanation and more precisely characterize when end-to-end can improve performance. When prediction targets are stochastic, a two-stage solution must make an a priori choice about which statistics of the target distribution to model -- we consider expectations over prediction targets -- while an end-to-end solution can make this choice adaptively. We show that the performance gap between a two-stage and end-to-end approach is closely related to the \emph{price of correlation} concept in stochastic optimization and show the implications of some existing POC results for our predict-then-optimize problem. We then consider a novel and particularly practical setting, where coefficients in the objective function depend on multiple prediction targets. We give explicit constructions where (1) two-stage performs unboundedly worse than end-to-end; and (2) two-stage is optimal. We identify a large set of real-world applications whose objective functions rely on multiple prediction targets but which nevertheless deploy two-stage solutions. We also use simulations to experimentally quantify performance gaps.

(2 min) We study how to evaluate the quantitative information content of a region within an image for a particular label. To this end, we bridge class activation maps with information theory. We develop an informative class activation map (infoCAM). Given a classification task, infoCAM depict how to accumulate information of partial regions to that of the entire image toward a label. Thus, we can utilise infoCAM to locate the most informative features for a label. When applied to an image classification task, infoCAM performs better than the traditional classification map in the weakly supervised object localisation task. We achieve state-of-the-art results on Tiny-ImageNet.

(2 min) Existing observational approaches for learning human preferences, such as inverse reinforcement learning, usually make strong assumptions about the observability of the human's environment. However, in reality, people make many important decisions under uncertainty. To better understand preference learning in these cases, we study the setting of inverse decision theory (IDT), a previously proposed framework where a human is observed making non-sequential binary decisions under uncertainty. In IDT, the human's preferences are conveyed through their loss function, which expresses a tradeoff between different types of mistakes. We give the first statistical analysis of IDT, providing conditions necessary to identify these preferences and characterizing the sample complexity -- the number of decisions that must be observed to learn the tradeoff the human is making to a desired precision. Interestingly, we show that it is actually easier to identify preferences when the decision problem is more uncertain. Furthermore, uncertain decision problems allow us to relax the unrealistic assumption that the human is an optimal decision maker but still identify their exact preferences; we give sample complexities in this suboptimal case as well. Our analysis contradicts the intuition that partial observability should make preference learning more difficult. It also provides a first step towards understanding and improving preference learning methods for uncertain and suboptimal humans.

(2 min) The capability of reinforcement learning (RL) agent directly depends on the diversity of learning scenarios the environment generates and how closely it captures real-world situations. However, existing environments/simulators lack the support to systematically model distributions over initial states and transition dynamics. Furthermore, in complex domains such as soccer, the space of possible scenarios is infinite, which makes it impossible for one research group to provide a comprehensive set of scenarios to train, test, and benchmark RL algorithms. To address this issue, for the first time, we adopt an existing formal scenario specification language, SCENIC, to intuitively model and generate interactive scenarios. We interfaced SCENIC to Google Research Soccer environment to create a platform called SCENIC4RL. Using this platform, we provide a dataset consisting of 36 scenario programs encoded in SCENIC and demonstration data generated from a subset of them. We share our experimental results to show the effectiveness of our dataset and the platform to train, test, and benchmark RL algorithms. More importantly, we open-source our platform to enable RL community to collectively contribute to constructing a comprehensive set of scenarios.

(2 min) Hoping to stimulate new research in individual animal identification from images, we propose to formulate the problem as the human-machine Continual Curation of images and animal identities. This is an open world recognition problem, where most new animals enter the system after its algorithms are initially trained and deployed. Continual Curation, as defined here, requires (1) an improvement in the effectiveness of current recognition methods, (2) a pairwise verification algorithm that allows the possibility of no decision, and (3) an algorithmic decision mechanism that seeks human input to guide the curation process. Error metrics must evaluate the ability of recognition algorithms to identify not only animals that have been seen just once or twice but also recognize new animals not in the database. An important measure of overall system performance is accuracy as a function of the amount of human input required.

(2 min) While the long-term effects of COVID-19 are yet to be determined, its immediate impact on crowdfunding is nonetheless significant. This study takes a computational approach to more deeply comprehend this change. Using a unique data set of all the campaigns published over the past two years on GoFundMe, we explore the factors that have led to the successful funding of a crowdfunding project. In particular, we study a corpus of crowdfunded projects, analyzing cover images and other variables commonly present on crowdfunding sites. Furthermore, we construct a classifier and a regression model to assess the significance of features based on XGBoost. In addition, we employ counterfactual analysis to investigate the causality between features and the success of crowdfunding. More importantly, sentiment analysis and the paired sample t-test are performed to examine the differences in crowdfunding campaigns before and after the COVID-19 outbreak that started in March 2020. First, we note that there is significant racial disparity in crowdfunding success. Second, we find that sad emotion expressed through the campaign's description became significant after the COVID-19 outbreak. Considering all these factors, our findings shed light on the impact of COVID-19 on crowdfunding campaigns.

(2 min) Previous studies dominantly target at self-supervised learning on real-valued networks and have achieved many promising results. However, on the more challenging binary neural networks (BNNs), this task has not yet been fully explored in the community. In this paper, we focus on this more difficult scenario: learning networks where both weights and activations are binary, meanwhile, without any human annotated labels. We observe that the commonly used contrastive objective is not satisfying on BNNs for competitive accuracy, since the backbone network contains relatively limited capacity and representation ability. Hence instead of directly applying existing self-supervised methods, which cause a severe decline in performance, we present a novel guided learning paradigm from real-valued to distill binary networks on the final prediction distribution, to minimize the loss and obtain desirable accuracy. Our proposed method can boost the simple contrastive learning baseline by an absolute gain of 5.5~15% on BNNs. We further reveal that it is difficult for BNNs to recover the similar predictive distributions as real-valued models when training without labels. Thus, how to calibrate them is key to address the degradation in performance. Extensive experiments are conducted on the large-scale ImageNet and downstream datasets. Our method achieves substantial improvement over the simple contrastive learning baseline, and is even comparable to many mainstream supervised BNN methods. Code is available at https://github.com/szq0214/S2-BNN.

(2 min) Recently, sound-based COVID-19 detection studies have shown great promise to achieve scalable and prompt digital pre-screening. However, there are still two unsolved issues hindering the practice. First, collected datasets for model training are often imbalanced, with a considerably smaller proportion of users tested positive, making it harder to learn representative and robust features. Second, deep learning models are generally overconfident in their predictions. Clinically, false predictions aggravate healthcare costs. Estimation of the uncertainty of screening would aid this. To handle these issues, we propose an ensemble framework where multiple deep learning models for sound-based COVID-19 detection are developed from different but balanced subsets from original data. As such, data are utilized more effectively compared to traditional up-sampling and down-sampling approaches: an AUC of 0.74 with a sensitivity of 0.68 and a specificity of 0.69 is achieved. Simultaneously, we estimate uncertainty from the disagreement across multiple models. It is shown that false predictions often yield higher uncertainty, enabling us to suggest the users with certainty higher than a threshold to repeat the audio test on their phones or to take clinical tests if digital diagnosis still fails. This study paves the way for a more robust sound-based COVID-19 automated screening system.

(2 min) Post-hoc gradient-based interpretability methods [Simonyan et al., 2013, Smilkov et al., 2017] that provide instance-specific explanations of model predictions are often based on assumption (A): magnitude of input gradients -- gradients of logits with respect to input -- noisily highlight discriminative task-relevant features. In this work, we test the validity of assumption (A) using a three-pronged approach. First, we develop an evaluation framework, DiffROAR, to test assumption (A) on four image classification benchmarks. Our results suggest that (i) input gradients of standard models (i.e., trained on original data) may grossly violate (A), whereas (ii) input gradients of adversarially robust models satisfy (A). Second, we then introduce BlockMNIST, an MNIST-based semi-real dataset, that by design encodes a priori knowledge of discriminative features. Our analysis on BlockMNIST leverages this information to validate as well as characterize differences between input gradient attributions of standard and robust models. Finally, we theoretically prove that our empirical findings hold on a simplified version of the BlockMNIST dataset. Specifically, we prove that input gradients of standard one-hidden-layer MLPs trained on this dataset do not highlight instance-specific signal coordinates, thus grossly violating assumption (A). Our findings motivate the need to formalize and test common assumptions in interpretability in a falsifiable manner [Leavitt and Morcos, 2020]. Additionally, we believe that the DiffROAR evaluation framework and BlockMNIST-based datasets can serve as sanity checks to audit instance-specific interpretability methods.

(2 min) Music source separation can be interpreted as the estimation of the constituent music sources that a music clip is composed of. In this work, we explore the single-channel singing voice separation problem from a multimodal perspective, by jointly learning from audio and visual modalities. To do so, we present Acappella, a dataset spanning around 46 hours of a cappella solo singing videos sourced from YouTube. We propose Y-Net, an audio-visual convolutional neural network which achieves state-of-the-art singing voice separation results on the Acappella dataset and compare it against its audio-only counterpart, U-Net, and a state-of-the-art audio-visual speech separation model. Singing voice separation can be particularly challenging when the audio mixture also comprises of other accompaniment voices and background sounds along with the target voice of interest. We demonstrate that our model can outperform the baseline models in the singing voice separation task in such challenging scenarios. The code, the pre-trained models and the dataset will be publicly available at https://ipcv.github.io/Acappella/

(2 min) In this paper, we explore the ability to model and infer personality types of opponents, predict their responses, and use this information to adapt a dialog agent's high-level strategy in negotiation tasks. Inspired by the idea of incorporating a theory of mind (ToM) into machines, we introduce a probabilistic formulation to encapsulate the opponent's personality type during both learning and inference. We test our approach on the CraigslistBargain dataset and show that our method using ToM inference achieves a 20% higher dialog agreement rate compared to baselines on a mixed population of opponents. We also find that our model displays diverse negotiation behavior with different types of opponents.

(2 min) Stochastic modelling is an essential component of the quantitative sciences, with hidden Markov models (HMMs) often playing a central role. Concurrently, the rise of quantum technologies promises a host of advantages in computational problems, typically in terms of the scaling of requisite resources such as time and memory. HMMs are no exception to this, with recent results highlighting quantum implementations of deterministic HMMs exhibiting superior memory and thermal efficiency relative to their classical counterparts. In many contexts however, non-deterministic HMMs are viable alternatives; compared to them the advantages of current quantum implementations do not always hold. Here, we provide a systematic prescription for constructing quantum implementations of non-deterministic HMMs that re-establish the quantum advantages against this broader class. Crucially, we show that whenever the classical implementation suffers from thermal dissipation due to its need to process information in a time-local manner, our quantum implementations will both mitigate some of this dissipation, and achieve an advantage in memory compression.

(2 min) Action-value estimation is a critical component of many reinforcement learning (RL) methods whereby sample complexity relies heavily on how fast a good estimator for action value can be learned. By viewing this problem through the lens of representation learning, good representations of both state and action can facilitate action-value estimation. While advances in deep learning have seamlessly driven progress in learning state representations, given the specificity of the notion of agency to RL, little attention has been paid to learning action representations. We conjecture that leveraging the combinatorial structure of multi-dimensional action spaces is a key ingredient for learning good representations of action. To test this, we set forth the action hypergraph networks framework -- a class of functions for learning action representations in multi-dimensional discrete action spaces with a structural inductive bias. Using this framework we realise an agent class based on a combination with deep Q-networks, which we dub hypergraph Q-networks. We show the effectiveness of our approach on a myriad of domains: illustrative prediction problems under minimal confounding effects, Atari 2600 games, and discretised physical control benchmarks.

(2 min) A video prediction model that generalizes to diverse scenes would enable intelligent agents such as robots to perform a variety of tasks via planning with the model. However, while existing video prediction models have produced promising results on small datasets, they suffer from severe underfitting when trained on large and diverse datasets. To address this underfitting challenge, we first observe that the ability to train larger video prediction models is often bottlenecked by the memory constraints of GPUs or TPUs. In parallel, deep hierarchical latent variable models can produce higher quality predictions by capturing the multi-level stochasticity of future observations, but end-to-end optimization of such models is notably difficult. Our key insight is that greedy and modular optimization of hierarchical autoencoders can simultaneously address both the memory constraints and the optimization challenges of large-scale video prediction. We introduce Greedy Hierarchical Variational Autoencoders (GHVAEs), a method that learns high-fidelity video predictions by greedily training each level of a hierarchical autoencoder. In comparison to state-of-the-art models, GHVAEs provide 17-55% gains in prediction performance on four video datasets, a 35-40% higher success rate on real robot tasks, and can improve performance monotonically by simply adding more modules.

(2 min) As passenger vehicle technologies have advanced, so have their capabilities to avoid obstacles, especially with developments in tires, suspensions, steering, as well as safety technologies like ABS, ESC, and more recently, ADAS systems. However, environments around passenger vehicles have also become more complex, and dangerous. There have previously been studies that outline driver tendencies and performance capabilities when attempting to avoid obstacles while driving passenger vehicles. Now that autonomous vehicles are being developed with obstacle avoidance capabilities, it is important to target performance that meets or exceeds that of human drivers. This manuscript highlights systems that are crucial for an emergency obstacle avoidance maneuver (EOAM) and identifies the state-of-the-art for each of the related systems, while considering the nuances of traveling at highway speeds. Some of the primary EOAM-related systems/areas that are discussed in this review are: general path planning methods, system hierarchies, decision-making, trajectory generation, and trajectory-tracking control methods. After concluding remarks, suggestions for future work which could lead to an ideal EOAM development, are discussed.

(2 min) Neural controlled differential equations (CDEs) are the continuous-time analogue of recurrent neural networks, as Neural ODEs are to residual networks, and offer a memory-efficient continuous-time way to model functions of potentially irregular time series. Existing methods for computing the forward pass of a Neural CDE involve embedding the incoming time series into path space, often via interpolation, and using evaluations of this path to drive the hidden state. Here, we use rough path theory to extend this formulation. Instead of directly embedding into path space, we instead represent the input signal over small time intervals through its \textit{log-signature}, which are statistics describing how the signal drives a CDE. This is the approach for solving \textit{rough differential equations} (RDEs), and correspondingly we describe our main contribution as the introduction of Neural RDEs. This extension has a purpose: by generalising the Neural CDE approach to a broader class of driving signals, we demonstrate particular advantages for tackling long time series. In this regime, we demonstrate efficacy on problems of length up to 17k observations and observe significant training speed-ups, improvements in model performance, and reduced memory requirements compared to existing approaches.

(2 min) Actors and critics in actor-critic reinforcement learning algorithms are functionally separate, yet they often use the same network architectures. This case study explores the performance impact of network sizes when considering actor and critic architectures independently. By relaxing the assumption of architectural symmetry, it is often possible for smaller actors to achieve comparable policy performance to their symmetric counterparts. Our experiments show up to 99% reduction in the number of network weights with an average reduction of 77% over multiple actor-critic algorithms on 9 independent tasks. Given that reducing actor complexity results in a direct reduction of run-time inference cost, we believe configurations of actors and critics are aspects of actor-critic design that deserve to be considered independently, particularly in resource-constrained applications or when deploying multiple actors simultaneously.

(2 min) Knowledge distillation is widely used as a means of improving the performance of a relatively simple student model using the predictions from a complex teacher model. Several works have shown that distillation significantly boosts the student's overall performance; however, are these gains uniform across all data subgroups? In this paper, we show that distillation can harm performance on certain subgroups, e.g., classes with few associated samples. We trace this behaviour to errors made by the teacher distribution being transferred to and amplified by the student model. To mitigate this problem, we present techniques which soften the teacher influence for subgroups where it is less reliable. Experiments on several image classification benchmarks show that these modifications of distillation maintain boost in overall accuracy, while additionally ensuring improvement in subgroup performance.

(2 min) We propose to learn a generative model via entropy interpolation with a Schr\"{o}dinger Bridge. The generative learning task can be formulated as interpolating between a reference distribution and a target distribution based on the Kullback-Leibler divergence. At the population level, this entropy interpolation is characterized via an SDE on $[0,1]$ with a time-varying drift term. At the sample level, we derive our Schr\"{o}dinger Bridge algorithm by plugging the drift term estimated by a deep score estimator and a deep density ratio estimator into the Euler-Maruyama method. Under some mild smoothness assumptions of the target distribution, we prove the consistency of both the score estimator and the density ratio estimator, and then establish the consistency of the proposed Schr\"{o}dinger Bridge approach. Our theoretical results guarantee that the distribution learned by our approach converges to the target distribution. Experimental results on multimodal synthetic data and benchmark data support our theoretical findings and indicate that the generative model via Schr\"{o}dinger Bridge is comparable with state-of-the-art GANs, suggesting a new formulation of generative learning. We demonstrate its usefulness in image interpolation and image inpainting.

(2 min) Tensor completion is the problem of estimating the missing values of high-order data from partially observed entries. Among several definitions of tensor rank, tensor ring rank affords the flexibility and accuracy needed to model tensors of different orders, which motivated recent efforts on tensor-ring completion. However, data corruption due to prevailing outliers poses major challenges to existing algorithms. In this paper, we develop a robust approach to tensor ring completion that uses an M-estimator as its error statistic, which can significantly alleviate the effect of outliers. Leveraging a half-quadratic (HQ) method, we reformulate the problem as one of weighted tensor completion. We present two HQ-based algorithms based on truncated singular value decomposition and matrix factorization along with their convergence and complexity analysis. Extendibility of the proposed approach to alternative definitions of tensor rank is also discussed. The experimental results demonstrate the superior performance of the proposed approach over state-of-the-art robust algorithms for tensor completion.

(2 min) In this work we present a novel solution for Video Instance Segmentation(VIS), that is automatically generating instance level segmentation masks along with object class and tracking them in a video. Our method improves the masks from segmentation and propagation branches in an online manner using the Mask Selection Network (MSN) hence limiting the noise accumulation during mask tracking. We propose an effective design of MSN by using patch-based convolutional neural network. The network is able to distinguish between very subtle differences between the masks and choose the better masks out of the associated masks accurately. Further, we make use of temporal consistency and process the video sequences in both forward and reverse manner as a post processing step to recover lost objects. The proposed method can be used to adapt any video object segmentation method for the task of VIS. Our method achieves a score of 49.1 mAP on 2021 YouTube-VIS Challenge and was ranked third place among more than 30 global teams. Our code will be available at https://github.com/SHI-Labs/Mask-Selection-Networks.

(2 min) The recently-introduced class of ordinary differential equation networks (ODE-Nets) establishes a fruitful connection between deep learning and dynamical systems. In this work, we reconsider formulations of the weights as continuous-depth functions using linear combinations of basis functions. This perspective allows us to compress the weights through a change of basis, without retraining, while maintaining near state-of-the-art performance. In turn, both inference time and the memory footprint are reduced, enabling quick and rigorous adaptation between computational environments. Furthermore, our framework enables meaningful continuous-in-time batch normalization layers using function projections. The performance of basis function compression is demonstrated by applying continuous-depth models to (a) image classification tasks using convolutional units and (b) sentence-tagging tasks using transformer encoder units.

(2 min) The intuition that local flatness of the loss landscape is correlated with better generalization for deep neural networks (DNNs) has been explored for decades, spawning many different flatness measures. Recently, this link with generalization has been called into question by a demonstration that many measures of flatness are vulnerable to parameter re-scaling which arbitrarily changes their value without changing neural network outputs. Here we show that, in addition, some popular variants of SGD such as Adam and Entropy-SGD, can also break the flatness-generalization correlation. As an alternative to flatness measures, we use a function based picture and propose using the log of Bayesian prior upon initialization, $\log P(f)$, as a predictor of the generalization when a DNN converges on function $f$ after training to zero error. The prior is directly proportional to the Bayesian posterior for functions that give zero error on a test set. For the case of image classification, we show that $\log P(f)$ is a significantly more robust predictor of generalization than flatness measures are. Whilst local flatness measures fail under parameter re-scaling, the prior/posterior, which is global quantity, remains invariant under re-scaling. Moreover, the correlation with generalization as a function of data complexity remains good for different variants of SGD.

(2 min) Entity matching (EM) refers to the problem of identifying tuple pairs in one or more relations that refer to the same real world entities. Supervised machine learning (ML) approaches, and deep learning based approaches in particular, typically achieve state-of-the-art matching results. However, these approaches require many labeled examples, in the form of matching and non-matching pairs, which are expensive and time-consuming to label. In this paper, we introduce Panda, a weakly supervised system specifically designed for EM. Panda uses the same labeling function abstraction as Snorkel, where labeling functions (LF) are user-provided programs that can generate large amounts of (somewhat noisy) labels quickly and cheaply, which can then be combined via a labeling model to generate accurate final predictions. To support users developing LFs for EM, Panda provides an integrated development environment (IDE) that lives in a modern browser architecture. Panda's IDE facilitates the development, debugging, and life-cycle management of LFs in the context of EM tasks, similar to how IDEs such as Visual Studio or Eclipse excel in general-purpose programming. Panda's IDE includes many novel features purpose-built for EM, such as smart data sampling, a builtin library of EM utility functions, automatically generated LFs, visual debugging of LFs, and finally, an EM-specific labeling model. We show in this demo that Panda IDE can greatly accelerate the development of high-quality EM solutions using weak supervision.

(2 min) First attempts of prediction of the facial growth (FG) direction were made over half of a century ago. Despite numerous attempts and elapsed time, a satisfactory method has not been established yet and the problem still poses a challenge for medical experts. To our knowledge, this paper is the first Machine Learning approach to the prediction of FG direction. Conducted data analysis reveals the inherent complexity of the problem and explains the reasons of difficulty in FG direction prediction based on 2D X-ray images. To perform growth forecasting, we employ a wide range of algorithms, from logistic regression, through tree ensembles to neural networks and consider three, slightly different, problem formulations. The resulting classification accuracy varies between 71% and 75%.

(2 min) Graphically-rich applications such as games are ubiquitous with attractive visual effects of Graphical User Interface (GUI) that offers a bridge between software applications and end-users. However, various types of graphical glitches may arise from such GUI complexity and have become one of the main component of software compatibility issues. Our study on bug reports from game development teams in NetEase Inc. indicates that graphical glitches frequently occur during the GUI rendering and severely degrade the quality of graphically-rich applications such as video games. Existing automated testing techniques for such applications focus mainly on generating various GUI test sequences and check whether the test sequences can cause crashes. These techniques require constant human attention to captures non-crashing bugs such as bugs causing graphical glitches. In this paper, we present the first step in automating the test oracle for detecting non-crashing bugs in graphically-rich applications. Specifically, we propose \texttt{GLIB} based on a code-based data augmentation technique to detect game GUI glitches. We perform an evaluation of \texttt{GLIB} on 20 real-world game apps (with bug reports available) and the result shows that \texttt{GLIB} can achieve 100\% precision and 99.5\% recall in detecting non-crashing bugs such as game GUI glitches. Practical application of \texttt{GLIB} on another 14 real-world games (without bug reports) further demonstrates that \texttt{GLIB} can effectively uncover GUI glitches, with 48 of 53 bugs reported by \texttt{GLIB} having been confirmed and fixed so far.

(2 min) Although deep neural networks (DNNs) have achieved a great success in various computer vision tasks, it is recently found that they are vulnerable to adversarial attacks. In this paper, we focus on the so-called \textit{backdoor attack}, which injects a backdoor trigger to a small portion of training data (also known as data poisoning) such that the trained DNN induces misclassification while facing examples with this trigger. To be specific, we carefully study the effect of both real and synthetic backdoor attacks on the internal response of vanilla and backdoored DNNs through the lens of Gard-CAM. Moreover, we show that the backdoor attack induces a significant bias in neuron activation in terms of the $\ell_\infty$ norm of an activation map compared to its $\ell_1$ and $\ell_2$ norm. Spurred by our results, we propose the \textit{$\ell_\infty$-based neuron pruning} to remove the backdoor from the backdoored DNN. Experiments show that our method could effectively decrease the attack success rate, and also hold a high classification accuracy for clean images.

(2 min) Learning generative probabilistic models is a core problem in machine learning, which presents significant challenges due to the curse of dimensionality. This paper proposes a joint dimensionality reduction and non-parametric density estimation framework, using a novel estimator that can explicitly capture the underlying distribution of appropriate reduced-dimension representations of the input data. The idea is to jointly design a nonlinear dimensionality reducing auto-encoder to model the training data in terms of a parsimonious set of latent random variables, and learn a canonical low-rank tensor model of the joint distribution of the latent variables in the Fourier domain. The proposed latent density model is non-parametric and universal, as opposed to the predefined prior that is assumed in variational auto-encoders. Joint optimization of the auto-encoder and the latent density estimator is pursued via a formulation which learns both by minimizing a combination of the negative log-likelihood in the latent domain and the auto-encoder reconstruction loss. We demonstrate that the proposed model achieves very promising results on toy, tabular, and image datasets on regression tasks, sampling, and anomaly detection.

(2 min) Despite the accomplishments of Generative Adversarial Networks (GANs) in modeling data distributions, training them remains a challenging task. A contributing factor to this difficulty is the non-intuitive nature of the GAN loss curves, which necessitates a subjective evaluation of the generated output to infer training progress. Recently, motivated by game theory, duality gap has been proposed as a domain agnostic measure to monitor GAN training. However, it is restricted to the setting when the GAN converges to a Nash equilibrium. But GANs need not always converge to a Nash equilibrium to model the data distribution. In this work, we extend the notion of duality gap to proximal duality gap that is applicable to the general context of training GANs where Nash equilibria may not exist. We show theoretically that the proximal duality gap is capable of monitoring the convergence of GANs to a wider spectrum of equilibria that subsumes Nash equilibria. We also theoretically establish the relationship between the proximal duality gap and the divergence between the real and generated data distributions for different GAN formulations. Our results provide new insights into the nature of GAN convergence. Finally, we validate experimentally the usefulness of proximal duality gap for monitoring and influencing GAN training.

(2 min) One of the main challenges in model-based reinforcement learning (RL) is to decide which aspects of the environment should be modeled. The value-equivalence (VE) principle proposes a simple answer to this question: a model should capture the aspects of the environment that are relevant for value-based planning. Technically, VE distinguishes models based on a set of policies and a set of functions: a model is said to be VE to the environment if the Bellman operators it induces for the policies yield the correct result when applied to the functions. As the number of policies and functions increase, the set of VE models shrinks, eventually collapsing to a single point corresponding to a perfect model. A fundamental question underlying the VE principle is thus how to select the smallest sets of policies and functions that are sufficient for planning. In this paper we take an important step towards answering this question. We start by generalizing the concept of VE to order-$k$ counterparts defined with respect to $k$ applications of the Bellman operator. This leads to a family of VE classes that increase in size as $k \rightarrow \infty$. In the limit, all functions become value functions, and we have a special instantiation of VE which we call proper VE or simply PVE. Unlike VE, the PVE class may contain multiple models even in the limit when all value functions are used. Crucially, all these models are sufficient for planning, meaning that they will yield an optimal policy despite the fact that they may ignore many aspects of the environment. We construct a loss function for learning PVE models and argue that popular algorithms such as MuZero and Muesli can be understood as minimizing an upper bound for this loss. We leverage this connection to propose a modification to MuZero and show that it can lead to improved performance in practice.

(2 min) Deep Reinforcement Learning (RL) techniques can benefit greatly from leveraging prior experience, which can be either self-generated or acquired from other entities. Action advising is a framework that provides a flexible way to transfer such knowledge in the form of actions between teacher-student peers. However, due to the realistic concerns, the number of these interactions is limited with a budget; therefore, it is crucial to perform these in the most appropriate moments. There have been several promising studies recently that address this problem setting especially from the student's perspective. Despite their success, they have some shortcomings when it comes to the practical applicability and integrity as an overall solution to the learning from advice challenge. In this paper, we extend the idea of advice reusing via teacher imitation to construct a unified approach that addresses both advice collection and advice utilisation problems. We also propose a method to automatically tune the relevant hyperparameters of these components on-the-fly to make it able to adapt to any task with minimal human intervention. The experiments we performed in $5$ different Atari games verify that our algorithm either surpasses or performs on-par with its top competitors while being far simpler to be employed. Furthermore, its individual components are also found to be providing significant advantages alone.

(2 min) With the rapid growth of renewable energy, lots of small photovoltaic (PV) prosumers emerge. Due to the uncertainty of solar power generation, there is a need for aggregated prosumers to predict solar power generation and whether solar power generation will be larger than load. This paper presents two interpretable neural networks to solve the problem: one binary classification neural network and one regression neural network. The neural networks are built using TensorFlow. The global feature importance and local feature contributions are examined by three gradient-based methods: Integrated Gradients, Expected Gradients, and DeepLIFT. Moreover, we detect abnormal cases when predictions might fail by estimating the prediction uncertainty using Bayesian neural networks. Neural networks, which are interpreted by gradient-based methods and complemented with uncertainty estimation, provide robust and explainable forecasting for decision-makers.

(2 min) Forecasting the number of Olympic medals for each nation is highly relevant for different stakeholders: Ex ante, sports betting companies can determine the odds while sponsors and media companies can allocate their resources to promising teams. Ex post, sports politicians and managers can benchmark the performance of their teams and evaluate the drivers of success. To significantly increase the Olympic medal forecasting accuracy, we apply machine learning, more specifically a two-staged Random Forest, thus outperforming more traditional na\"ive forecast for three previous Olympics held between 2008 and 2016 for the first time. Regarding the Tokyo 2020 Games in 2021, our model suggests that the United States will lead the Olympic medal table, winning 120 medals, followed by China (87) and Great Britain (74). Intriguingly, we predict that the current COVID-19 pandemic will not significantly alter the medal count as all countries suffer from the pandemic to some extent (data inherent) and limited historical data points on comparable diseases (model inherent).

(2 min) The discovery of causal relationships is a fundamental problem in science and medicine. In recent years, many elegant approaches to discovering causal relationships between two variables from observational data have been proposed. However, most of these deal only with purely directed causal relationships and cannot detect latent common causes. Here, we devise a general heuristic which takes a causal discovery algorithm that can only distinguish purely directed causal relations and modifies it to also detect latent common causes. We apply our method to two directed causal discovery algorithms, the Information Geometric Causal Inference of (Daniusis et al., 2010) and the Kernel Conditional Deviance for Causal Inference of (Mitrovic, Sejdinovic, & Teh, 2018), and extensively test on synthetic data -- detecting latent common causes in additive, multiplicative and complex noise regimes -- and on real data, where we are able to detect known common causes. In addition to detecting latent common causes, our experiments demonstrate that both the modified algorithms preserve the performance of the original in distinguishing directed causal relations.

(2 min) This paper serves as a survey of recent advances in large margin training and its theoretical foundations, mostly for (nonlinear) deep neural networks (DNNs) that are probably the most prominent machine learning models for large-scale data in the community over the past decade. We generalize the formulation of classification margins from classical research to latest DNNs, summarize theoretical connections between the margin, network generalization, and robustness, and introduce recent efforts in enlarging the margins for DNNs comprehensively. Since the viewpoint of different methods is discrepant, we categorize them into groups for ease of comparison and discussion in the paper. Hopefully, our discussions and overview inspire new research work in the community that aim to improve the performance of DNNs, and we also point to directions where the large margin principle can be verified to provide theoretical evidence why certain regularizations for DNNs function well in practice. We managed to shorten the paper such that the crucial spirit of large margin learning and related methods are better emphasized.

(2 min) We study the differential properties of higher-order statistical probabilistic programs with recursion and conditioning. Our starting point is an open problem posed by Hongseok Yang: what class of statistical probabilistic programs have densities that are differentiable almost everywhere? To formalise the problem, we consider Statistical PCF (SPCF), an extension of call-by-value PCF with real numbers, and constructs for sampling and conditioning. We give SPCF a sampling-style operational semantics a la Borgstrom et al., and study the associated weight (commonly referred to as the density) function and value function on the set of possible execution traces. Our main result is that almost-surely terminating SPCF programs, generated from a set of primitive functions (e.g. the set of analytic functions) satisfying mild closure properties, have weight and value functions that are almost-everywhere differentiable. We use a stochastic form of symbolic execution to reason about almost-everywhere differentiability. A by-product of this work is that almost-surely terminating deterministic (S)PCF programs with real parameters denote functions that are almost-everywhere differentiable. Our result is of practical interest, as almost-everywhere differentiability of the density function is required to hold for the correctness of major gradient-based inference algorithms.

(2 min) Modeling and simulating a power distribution network (PDN) for printed circuit boards (PCBs) with irregular board shapes and multi-layer stackup is computationally inefficient using full-wave simulations. This paper presents a new concept of using deep learning for PDN impedance prediction. A boundary element method (BEM) is applied to efficiently calculate the impedance for arbitrary board shape and stackup. Then over one million boards with different shapes, stackup, IC location, and decap placement are randomly generated to train a deep neural network (DNN). The trained DNN can predict the impedance accurately for new board configurations that have not been used for training. The consumed time using the trained DNN is only 0.1 seconds, which is over 100 times faster than the BEM method and 5000 times faster than full-wave simulations.

(2 min) Super-resolution (SR) is a one-to-many task with multiple possible solutions. However, previous works were not concerned about this characteristic. For a one-to-many pipeline, the generator should be able to generate multiple estimates of the reconstruction, and not be penalized for generating similar and equally realistic images. To achieve this, we propose adding weighted pixel-wise noise after every Residual-in-Residual Dense Block (RRDB) to enable the generator to generate various images. We modify the strict content loss to not penalize the stochastic variation in reconstructed images as long as it has consistent content. Additionally, we observe that there are out-of-focus regions in the DIV2K, DIV8K datasets that provide unhelpful guidelines. We filter blurry regions in the training data using the method of [10]. Finally, we modify the discriminator to receive the low-resolution image as a reference image along with the target image to provide better feedback to the generator. Using our proposed methods, we were able to improve the performance of ESRGAN in x4 perceptual SR and achieve the state-of-the-art LPIPS score in x16 perceptual extreme SR.

(2 min) This paper presents TS2Vec, a universal framework for learning timestamp-level representations of time series. Unlike existing methods, TS2Vec performs timestamp-wise discrimination, which learns a contextual representation vector directly for each timestamp. We find that the learned representations have superior predictive ability. A linear regression trained on top of the learned representations outperforms previous SOTAs for supervised time series forecasting. Also, the instance-level representations can be simply obtained by applying a max pooling layer on top of learned representations of all timestamps. We conduct extensive experiments on time series classification tasks to evaluate the quality of instance-level representations. As a result, TS2Vec achieves significant improvement compared with existing SOTAs of unsupervised time series representation on 125 UCR datasets and 29 UEA datasets. The source code is publicly available at https://github.com/yuezhihan/ts2vec.

(2 min) In this paper, we introduce a new acoustic leakage dataset of gas pipelines, called as GPLA-12, which has 12 categories over 684 training/testing acoustic signals. Unlike massive image and voice datasets, there have relatively few acoustic signal datasets, especially for engineering fault detection. In order to enhance the development of fault diagnosis, we collect acoustic leakage signals on the basis of an intact gas pipe system with external artificial leakages, and then preprocess the collected data with structured tailoring which are turned into GPLA-12. GPLA-12 dedicates to serve as a feature learning dataset for time-series tasks and classifications. To further understand the dataset, we train both shadow and deep learning algorithms to observe the performance. The dataset as well as the pretrained models have been released at both www.daip.club and github.com/Deep-AI-Application-DAIP

(2 min) Few-shot dataset generalization is a challenging variant of the well-studied few-shot classification problem where a diverse training set of several datasets is given, for the purpose of training an adaptable model that can then learn classes from new datasets using only a few examples. To this end, we propose to utilize the diverse training set to construct a universal template: a partial model that can define a wide array of dataset-specialized models, by plugging in appropriate components. For each new few-shot classification problem, our approach therefore only requires inferring a small number of parameters to insert into the universal template. We design a separate network that produces an initialization of those parameters for each given task, and we then fine-tune its proposed initialization via a few steps of gradient descent. Our approach is more parameter-efficient, scalable and adaptable compared to previous methods, and achieves the state-of-the-art on the challenging Meta-Dataset benchmark.

(2 min) The popularity of Internet-of-Things (IoT) has provided us with unprecedented opportunities to enable a variety of emerging services in a smart home environment. Among those services, sensing the liquid level in a container is critical to building many smart home and mobile healthcare applications that improve the quality of life. This paper presents LiquidSense, a liquid-level sensing system that is low-cost, high accuracy, widely applicable to different daily liquids and containers, and can be easily integrated with existing smart home networks. LiquidSense uses an existing home WiFi network and a low-cost transducer that attached to the container to sense the resonance of the container for liquid level detection. In particular, our system mounts a low-cost transducer on the surface of the container and emits a well-designed chirp signal to make the container resonant, which introduces subtle changes to the home WiFi signals. By analyzing the subtle phase changes of the WiFi signals, LiquidSense extracts the resonance frequency as a feature for liquid level detection. Our system constructs prediction models for both continuous and discrete predictions using curve fitting and SVM respectively. We evaluate LiquidSense in home environments with containers of three different materials and six types of liquids. Results show that LiquidSense achieves an overall accuracy of 97% for continuous prediction and an overall F-score of 0.968 for discrete prediction. Results also show that our system has a large coverage in a home environment and works well under non-line-of-sight (NLOS) scenarios.

(2 min) End-to-end speech translation models have become a new trend in research due to their potential of reducing error propagation. However, these models still suffer from the challenge of data scarcity. How to effectively use unlabeled or other parallel corpora from machine translation is promising but still an open problem. In this paper, we propose Cross Speech-Text Network (XSTNet), an end-to-end model for speech-to-text translation. XSTNet takes both speech and text as input and outputs both transcription and translation text. The model benefits from its three key design aspects: a self-supervised pre-trained sub-network as the audio encoder, a multi-task training objective to exploit additional parallel bilingual text, and a progressive training procedure. We evaluate the performance of XSTNet and baselines on the MuST-C En-X and LibriSpeech En-Fr datasets. In particular, XSTNet achieves state-of-the-art results on all language directions with an average BLEU of 28.8, outperforming the previous best method by 3.2 BLEU. Code, models, cases, and more detailed analysis are available at https://github.com/ReneeYe/XSTNet.

(2 min) Hate speech and profanity detection suffer from data sparsity, especially for languages other than English, due to the subjective nature of the tasks and the resulting annotation incompatibility of existing corpora. In this study, we identify profane subspaces in word and sentence representations and explore their generalization capability on a variety of similar and distant target tasks in a zero-shot setting. This is done monolingually (German) and cross-lingually to closely-related (English), distantly-related (French) and non-related (Arabic) tasks. We observe that, on both similar and distant target tasks and across all languages, the subspace-based representations transfer more effectively than standard BERT representations in the zero-shot setting, with improvements between F1 +10.9 and F1 +42.9 over the baselines across all tested monolingual and cross-lingual scenarios.

(2 min) Query by Example is a well-known information retrieval task in which a document is chosen by the user as the search query and the goal is to retrieve relevant documents from a large collection. However, a document often covers multiple aspects of a topic. To address this scenario we introduce the task of faceted Query by Example in which users can also specify a finer grained aspect in addition to the input query document. We focus on the application of this task in scientific literature search. We envision models which are able to retrieve scientific papers analogous to a query scientific paper along specifically chosen rhetorical structure elements as one solution to this problem. In this work, the rhetorical structure elements, which we refer to as facets, indicate backgrounds, methods, or results of a scientific paper. We introduce and describe an expert annotated test collection to evaluate models trained to perform this task. Our test collection consists of a diverse set of 50 query documents, drawn from computational linguistics and machine learning venues. We carefully followed the annotation guideline used by TREC for depth-k pooling (k = 100 or 250) and the resulting data collection consists of graded relevance scores with high annotation agreement. The data is freely available for research purposes.

(2 min) One of the key problems in multi-label text classification is how to take advantage of the correlation among labels. However, it is very challenging to directly model the correlations among labels in a complex and unknown label space. In this paper, we propose a Label Mask multi-label text classification model (LM-MTC), which is inspired by the idea of cloze questions of language model. LM-MTC is able to capture implicit relationships among labels through the powerful ability of pre-train language models. On the basis, we assign a different token to each potential label, and randomly mask the token with a certain probability to build a label based Masked Language Model (MLM). We train the MTC and MLM together, further improving the generalization ability of the model. A large number of experiments on multiple datasets demonstrate the effectiveness of our method.

(2 min) Predicting gender by the name is not a simple task. In many applications, especially in the natural language processing (NLP) field, this task may be necessary, mainly when considering foreign names. Some machine learning algorithms can satisfactorily perform the prediction. In this paper, we examined and implemented feedforward and recurrent deep neural network models, such as MLP, RNN, GRU, CNN, and BiLSTM, to classify gender through the first name. A dataset of Brazilian names is used to train and evaluate the models. We analyzed the accuracy, recall, precision, and confusion matrix to measure the models' performances. The results indicate that the gender prediction can be performed from the feature extraction strategy looking at the names as a set of strings. Some models accurately predict the gender in more than 90% of the cases. The recurrent models overcome the feedforward models in this binary classification problem.

(2 min) Punctuation prediction for automatic speech recognition (ASR) output transcripts plays a crucial role for improving the readability of the ASR transcripts and for improving the performance of downstream natural language processing applications. However, achieving good performance on punctuation prediction often requires large amounts of labeled speech transcripts, which is expensive and laborious. In this paper, we propose a Discriminative Self-Training approach with weighted loss and discriminative label smoothing to exploit unlabeled speech transcripts. Experimental results on the English IWSLT2011 benchmark test set and an internal Chinese spoken language dataset demonstrate that the proposed approach achieves significant improvement on punctuation prediction accuracy over strong baselines including BERT, RoBERTa, and ELECTRA models. The proposed Discriminative Self-Training approach outperforms the vanilla self-training approach. We establish a new state-of-the-art (SOTA) on the IWSLT2011 test set, outperforming the current SOTA model by 1.3% absolute gain on F$_1$.

(2 min) In the traditional cascading architecture for spoken language understanding (SLU), it has been observed that automatic speech recognition errors could be detrimental to the performance of natural language understanding. End-to-end (E2E) SLU models have been proposed to directly map speech input to desired semantic frame with a single model, hence mitigating ASR error propagation. Recently, pre-training technologies have been explored for these E2E models. In this paper, we propose a novel joint textual-phonetic pre-training approach for learning spoken language representations, aiming at exploring the full potentials of phonetic information to improve SLU robustness to ASR errors. We explore phoneme labels as high-level speech features, and design and compare pre-training tasks based on conditional masked language model objectives and inter-sentence relation objectives. We also investigate the efficacy of combining textual and phonetic information during fine-tuning. Experimental results on spoken language understanding benchmarks, Fluent Speech Commands and SNIPS, show that the proposed approach significantly outperforms strong baseline models and improves robustness of spoken language understanding to ASR errors.

(2 min) In recent years, great success has been achieved in the field of natural language processing (NLP), thanks in part to the considerable amount of annotated resources. For named entity recognition (NER), most languages do not have such an abundance of labeled data as English, so the performances of those languages are relatively lower. To improve the performance, we propose a general approach called Back Attention Network (BAN). BAN uses a translation system to translate other language sentences into English and then applies a new mechanism named back attention knowledge transfer to obtain task-specific information from pre-trained high-resource languages NER model. This strategy can transfer high-layer features of well-trained model and enrich the semantic representations of the original language. Experiments on three different language datasets indicate that the proposed approach outperforms other state-of-the-art methods.

(2 min) In multi-hop QA, answering complex questions entails iterative document retrieval for finding the missing entity of the question. The main steps of this process are sub-question detection, document retrieval for the sub-question, and generation of a new query for the final document retrieval. However, building a dataset that contains complex questions with sub-questions and their corresponding documents requires costly human annotation. To address the issue, we propose a new method for weakly supervised multi-hop retriever pre-training without human efforts. Our method includes 1) a pre-training task for generating vector representations of complex questions, 2) a scalable data generation method that produces the nested structure of question and sub-question as weak supervision for pre-training, and 3) a pre-training model structure based on dense encoders. We conduct experiments to compare the performance of our pre-trained retriever with several state-of-the-art models on end-to-end multi-hop QA as well as document retrieval. The experimental results show that our pre-trained retriever is effective and also robust on limited data and computational resources.

(2 min) Human creativity generates novel ideas to solve real-world problems. This thereby grants us the power to transform the surrounding world and extend our human attributes beyond what is currently possible. Creative ideas are not just new and unexpected, but are also successful in providing solutions that are useful, efficient and valuable. Thus, creativity optimizes the use of available resources and increases wealth. The origin of human creativity, however, is poorly understood, and semantic measures that could predict the success of generated ideas are currently unknown. Here, we analyze a dataset of design problem-solving conversations in real-world settings by using 49 semantic measures based on WordNet 3.1 and demonstrate that a divergence of semantic similarity, an increased information content, and a decreased polysemy predict the success of generated ideas. The first feedback from clients also enhances information content and leads to a divergence of successful ideas in creative problem solving. These results advance cognitive science by identifying real-world processes in human problem solving that are relevant to the success of produced solutions and provide tools for real-time monitoring of problem solving, student training and skill acquisition. A selected subset of information content (IC S\'anchez-Batet) and semantic similarity (Lin/S\'anchez-Batet) measures, which are both statistically powerful and computationally fast, could support the development of technologies for computer-assisted enhancements of human creativity or for the implementation of creativity in machines endowed with general artificial intelligence.

(2 min) Generating synthesised singing voice with models trained on speech data has many advantages due to the models' flexibility and controllability. However, since the information about the temporal relationship between segments and beats are lacking in speech training data, the synthesised singing may sound off-beat at times. Therefore, the availability of the information on the temporal relationship between speech segments and music beats is crucial. The current study investigated the segment-beat synchronisation in singing data, with hypotheses formed based on the linguistics theories of P-centre and sonority hierarchy. A Mandarin corpus and an English corpus of professional singing data were manually annotated and analysed. The results showed that the presence of musical beats was more dependent on segment duration than sonority. However, the sonority hierarchy and the P-centre theory were highly related to the location of beats. Mandarin and English demonstrated cross-linguistic variations despite exhibiting common patterns.

(2 min) While current state-of-the-art Automatic Speech Recognition (ASR) systems achieve high accuracy on typical speech, they suffer from significant performance degradation on disordered speech and other atypical speech patterns. Personalization of ASR models, a commonly applied solution to this problem, is usually performed in a server-based training environment posing problems around data privacy, delayed model-update times, and communication cost for copying data and models between mobile device and server infrastructure. In this paper, we present an approach to on-device based ASR personalization with very small amounts of speaker-specific data. We test our approach on a diverse set of 100 speakers with disordered speech and find median relative word error rate improvement of 71% with only 50 short utterances required per speaker. When tested on a voice-controlled home automation platform, on-device personalized models show a median task success rate of 81%, compared to only 40% of the unadapted models.

(2 min) Code-mixing is a frequent communication style among multilingual speakers where they mix words and phrases from two different languages in the same utterance of text or speech. Identifying and filtering code-mixed text is a challenging task due to its co-existence with monolingual and noisy text. Over the years, several code-mixing metrics have been extensively used to identify and validate code-mixed text quality. This paper demonstrates several inherent limitations of code-mixing metrics with examples from the already existing datasets that are popularly used across various experiments.

(2 min) By implicitly recognizing a user based on his/her speech input, speaker identification enables many downstream applications, such as personalized system behavior and expedited shopping checkouts. Based on whether the speech content is constrained or not, both text-dependent (TD) and text-independent (TI) speaker recognition models may be used. We wish to combine the advantages of both types of models through an ensemble system to make more reliable predictions. However, any such combined approach has to be robust to incomplete inputs, i.e., when either TD or TI input is missing. As a solution we propose a fusion of embeddings network foenet architecture, combining joint learning with neural attention. We compare foenet with four competitive baseline methods on a dataset of voice assistant inputs, and show that it achieves higher accuracy than the baseline and score fusion methods, especially in the presence of incomplete inputs.

(2 min) Digital assistants are experiencing rapid growth due to their ability to assist users with day-to-day tasks where most dialogues are happening multi-turn. However, evaluating multi-turn dialogues remains challenging, especially at scale. We suggest a context-sensitive method to estimate the turn-level satisfaction for dialogue considering various types of user preferences. The costs of interactions between users and dialogue systems are formulated using a budget consumption concept. We assume users have an initial interaction budget for a dialogue formed based on the task complexity and that each turn has a cost. When the task is completed, or the budget has been exhausted, users quit the dialogue. We demonstrate our method's effectiveness by extensive experimentation with a simulated dialogue platform and real multi-turn dialogues.

(2 min) We show that with small-to-medium training data, fine-tuning only the bias terms (or a subset of the bias terms) of pre-trained BERT models is competitive with (and sometimes better than) fine-tuning the entire model. For larger data, bias-only fine-tuning is competitive with other sparse fine-tuning methods. Besides their practical utility, these findings are relevant for the question of understanding the commonly-used process of finetuning: they support the hypothesis that finetuning is mainly about exposing knowledge induced by language-modeling training, rather than learning new task-specific linguistic knowledge.

(2 min) Sentiment analysis as a sub-field of natural language processing has received increased attention in the past decade enabling organisations to more effectively manage their reputation through online media monitoring. Many drivers impact reputation, however, this thesis focuses only the aspect of financial performance and explores the gap with regards to financial sentiment analysis in a South African context. Results showed that pre-trained sentiment analysers are least effective for this task and that traditional lexicon-based and machine learning approaches are best suited to predict financial sentiment of news articles. The evaluated methods produced accuracies of 84\%-94\%. The predicted sentiments correlated quite well with share price and highlighted the potential use of sentiment as an indicator of financial performance. A main contribution of the study was updating an existing sentiment dictionary for financial sentiment analysis. Model generalisation was less acceptable due to the limited amount of training data used. Future work includes expanding the data set to improve general usability and contribute to an open-source financial sentiment analyser for South African data.

(2 min) In this paper, we propose a novel edge-editing approach to extract relation information from a document. We treat the relations in a document as a relation graph among entities in this approach. The relation graph is iteratively constructed by editing edges of an initial graph, which might be a graph extracted by another system or an empty graph. The way to edit edges is to classify them in a close-first manner using the document and temporally-constructed graph information; each edge is represented with a document context information by a pretrained transformer model and a graph context information by a graph convolutional neural network model. We evaluate our approach on the task to extract material synthesis procedures from materials science texts. The experimental results show the effectiveness of our approach in editing the graphs initialized by our in-house rule-based system and empty graphs.

(2 min) In deep neural network modeling, the most common practice is to stack a number of recurrent, convolutional, or feed-forward layers in order to obtain high-quality continuous space representations which in turn improves the quality of the network's prediction. Conventionally, each layer in the stack has its own parameters which leads to a significant increase in the number of model parameters. In this paper, we propose to share parameters across all layers thereby leading to a recurrently stacked neural network model. We report on an extensive case study on neural machine translation (NMT), where we apply our proposed method to an encoder-decoder based neural network model, i.e., the Transformer model, and experiment with three Japanese--English translation datasets. We empirically demonstrate that the translation quality of a model that recurrently stacks a single layer 6 times, despite having significantly fewer parameters, approaches that of a model that stacks 6 layers where each layer has different parameters. We also explore the limits of recurrent stacking where we train extremely deep NMT models. This paper also examines the utility of our recurrently stacked model as a student model through transfer learning via leveraging pre-trained parameters and knowledge distillation, and shows that it compensates for the performance drops in translation quality that the direct training of recurrently stacked model brings. We also show how transfer learning helps in faster decoding on top of the already reduced number of parameters due to recurrent stacking. Finally, we analyze the effects of recurrently stacked layers by visualizing the attentions of models that use recurrently stacked layers and models that do not.

(2 min) Noise contrastive learning is a popular technique for unsupervised representation learning. In this approach, a representation is obtained via reduction to supervised learning, where given a notion of semantic similarity, the learner tries to distinguish a similar (positive) example from a collection of random (negative) examples. The success of modern contrastive learning pipelines relies on many parameters such as the choice of data augmentation, the number of negative examples, and the batch size; however, there is limited understanding as to how these parameters interact and affect downstream performance. We focus on disambiguating the role of one of these parameters: the number of negative examples. Theoretically, we show the existence of a collision-coverage trade-off suggesting that the optimal number of negative examples should scale with the number of underlying concepts in the data. Empirically, we scrutinize the role of the number of negatives in both NLP and vision tasks. In the NLP task, we find that the results broadly agree with our theory, while our vision experiments are murkier with performance sometimes even being insensitive to the number of negatives. We discuss plausible explanations for this behavior and suggest future directions to better align theory and practice.

(2 min) Quotations are crucial for successful explanations and persuasions in interpersonal communications. However, finding what to quote in a conversation is challenging for both humans and machines. This work studies automatic quotation generation in an online conversation and explores how language consistency affects whether a quotation fits the given context. Here, we capture the contextual consistency of a quotation in terms of latent topics, interactions with the dialogue history, and coherence to the query turn's existing content. Further, an encoder-decoder neural framework is employed to continue the context with a quotation via language generation. Experiment results on two large-scale datasets in English and Chinese demonstrate that our quotation generation model outperforms the state-of-the-art models. Further analysis shows that topic, interaction, and query consistency are all helpful to learn how to quote in online conversations.

(2 min) Automatic speech recognition (ASR) models make fewer errors when more surrounding speech information is presented as context. Unfortunately, acquiring a larger future context leads to higher latency. There exists an inevitable trade-off between speed and accuracy. Naively, to fit different latency requirements, people have to store multiple models and pick the best one under the constraints. Instead, a more desirable approach is to have a single model that can dynamically adjust its latency based on different constraints, which we refer to as Multi-mode ASR. A Multi-mode ASR model can fulfill various latency requirements during inference -- when a larger latency becomes acceptable, the model can process longer future context to achieve higher accuracy and when a latency budget is not flexible, the model can be less dependent on future context but still achieve reliable accuracy. In pursuit of Multi-mode ASR, we propose Stochastic Future Context, a simple training procedure that samples one streaming configuration in each iteration. Through extensive experiments on AISHELL-1 and LibriSpeech datasets, we show that a Multi-mode ASR model rivals, if not surpasses, a set of competitive streaming baselines trained with different latency budgets.

(2 min) Joint extraction of entities and relations from unstructured texts is a crucial task in information extraction. Recent methods achieve considerable performance but still suffer from some inherent limitations, such as redundancy of relation prediction, poor generalization of span-based extraction and inefficiency. In this paper, we decompose this task into three subtasks, Relation Judgement, Entity Extraction and Subject-object Alignment from a novel perspective and then propose a joint relational triple extraction framework based on Potential Relation and Global Correspondence (PRGC). Specifically, we design a component to predict potential relations, which constrains the following entity extraction to the predicted relation subset rather than all relations; then a relation-specific sequence tagging component is applied to handle the overlapping problem between subjects and objects; finally, a global correspondence component is designed to align the subject and object into a triple with low-complexity. Extensive experiments show that PRGC achieves state-of-the-art performance on public benchmarks with higher efficiency and delivers consistent performance gain on complex scenarios of overlapping triples.

(2 min) Entity linking (EL), the task of disambiguating mentions in text by linking them to entities in a knowledge graph, is crucial for text understanding, question answering or conversational systems. Entity linking on short text (e.g., single sentence or question) poses particular challenges due to limited context. While prior approaches use either heuristics or black-box neural methods, here we propose LNN-EL, a neuro-symbolic approach that combines the advantages of using interpretable rules based on first-order logic with the performance of neural learning. Even though constrained to using rules, LNN-EL performs competitively against SotA black-box neural approaches, with the added benefits of extensibility and transferability. In particular, we show that we can easily blend existing rule templates given by a human expert, with multiple types of features (priors, BERT encodings, box embeddings, etc), and even scores resulting from previous EL methods, thus improving on such methods. For instance, on the LC-QuAD-1.0 dataset, we show more than $4$\% increase in F1 score over previous SotA. Finally, we show that the inductive bias offered by using logic results in learned rules that transfer well across datasets, even without fine tuning, while maintaining high accuracy.

(2 min) Detecting what emotions are expressed in text is a well-studied problem in natural language processing. However, research on finer grained emotion analysis such as what causes an emotion is still in its infancy. We present solutions that tackle both emotion recognition and emotion cause detection in a joint fashion. Considering that common-sense knowledge plays an important role in understanding implicitly expressed emotions and the reasons for those emotions, we propose novel methods that combine common-sense knowledge via adapted knowledge models with multi-task learning to perform joint emotion classification and emotion cause tagging. We show performance improvement on both tasks when including common-sense reasoning and a multitask framework. We provide a thorough analysis to gain insights into model performance.

(2 min) Due to the subjectivity of the summarization, it is a good practice to have more than one gold summary for each training document. However, many modern large-scale abstractive summarization datasets have only one-to-one samples written by different human with different styles. The impact of this phenomenon is understudied. We formulate the differences among possible multiple expressions summarizing the same content as subjective bias and examine the role of this bias in the context of abstractive summarization. In this paper a lightweight and effective method to extract the feature embeddings of subjective styles is proposed. Results of summarization models trained on style-clustered datasets show that there are certain types of styles that lead to better convergence, abstraction and generalization. The reproducible code and generated summaries are available online.

(2 min) Public concern detection provides potential guidance to the authorities for crisis management before or during a pandemic outbreak. Detecting people's concerns and attention from online social media platforms has been widely acknowledged as an effective approach to relieve public panic and prevent a social crisis. However, detecting concerns in time from massive information in social media turns out to be a big challenge, especially when sufficient manually labeled data is in the absence of public health emergencies, e.g., COVID-19. In this paper, we propose a novel end-to-end deep learning model to identify people's concerns and the corresponding relations based on Graph Convolutional Network and Bi-directional Long Short Term Memory integrated with Concern Graph. Except for the sequential features from BERT embeddings, the regional features of tweets can be extracted by the Concern Graph module, which not only benefits the concern detection but also enables our model to be high noise-tolerant. Thus, our model can address the issue of insufficient manually labeled data. We conduct extensive experiments to evaluate the proposed model by using both manually labeled tweets and automatically labeled tweets. The experimental results show that our model can outperform the state-of-art models on real-world datasets.

(2 min) Building a benchmark dataset for hate speech detection presents several challenges. Firstly, because hate speech is relatively rare -- e.g., less than 3\% of Twitter posts are hateful \citep{founta2018large} -- random sampling of tweets to annotate is inefficient in capturing hate speech. A common practice is to only annotate tweets containing known ``hate words'', but this risks yielding a biased benchmark that only partially captures the real-world phenomenon of interest. A second challenge is that definitions of hate speech tend to be highly variable and subjective. Annotators having diverse prior notions of hate speech may not only disagree with one another but also struggle to conform to specified labeling guidelines. Our key insight is that the rarity and subjectivity of hate speech are akin to that of relevance in information retrieval (IR). This connection suggests that well-established methodologies for creating IR test collections might also be usefully applied to create better benchmark datasets for hate speech detection. Firstly, to intelligently and efficiently select which tweets to annotate, we apply established IR techniques of {\em pooling} and {\em active learning}. Secondly, to improve both consistency and value of annotations, we apply {\em task decomposition} \cite{Zhang-sigir14} and {\em annotator rationale} \cite{mcdonnell16-hcomp} techniques. Using the above techniques, we create and share a new benchmark dataset\footnote{We will release the dataset upon publication.} for hate speech detection with broader coverage than prior datasets. We also show a dramatic drop in accuracy of existing detection models when tested on these broader forms of hate. Collected annotator rationales not only provide documented support for labeling decisions but also create exciting future work opportunities for dual-supervision and/or explanation generation in modeling.

(2 min) In this paper, we present GEM as a General Evaluation benchmark for Multimodal tasks. Different from existing datasets such as GLUE, SuperGLUE, XGLUE and XTREME that mainly focus on natural language tasks, GEM is a large-scale vision-language benchmark, which consists of GEM-I for image-language tasks and GEM-V for video-language tasks. Comparing with existing multimodal datasets such as MSCOCO and Flicker30K for image-language tasks, YouCook2 and MSR-VTT for video-language tasks, GEM is not only the largest vision-language dataset covering image-language tasks and video-language tasks at the same time, but also labeled in multiple languages. We also provide two baseline models for this benchmark. We will release the dataset, code and baseline models, aiming to advance the development of multilingual multimodal research.

(2 min) We aim to create an unprecedented attempt to build an end-to-end Question Answering (QA) over Knowledge Graphs (KGs), which can construct SPARQL queries from natural language questions and generate a verbalized answer to its queries. Hence, we introduce SPBERT, a Transformer-based language model pre-trained on massive SPARQL query logs. By incorporating masked language modelling objective and word structural objective, SPBERT can learn general-purpose representations in both natural language and SPARQL query language and make the most of the sequential order of words that are crucial for structured language like SPARQL. In this paper, we investigate how SPBERT and encoder-decoder architecture can be adapted for Knowledge-based QA corpora. We conduct exhaustive experiments on two auxiliary tasks, including SPARQL Query Construction and Answer Verbalization Generation. Results show that SPBERT obtains promising performance and achieves state-of-the-art results on several of these tasks.

(2 min) Several years of research have shown that machine-learning systems are vulnerable to adversarial examples, both in theory and in practice. Until now, such attacks have primarily targeted visual models, exploiting the gap between human and machine perception. Although text-based models have also been attacked with adversarial examples, such attacks struggled to preserve semantic meaning and indistinguishability. In this paper, we explore a large class of adversarial examples that can be used to attack text-based models in a black-box setting without making any human-perceptible visual modification to inputs. We use encoding-specific perturbations that are imperceptible to the human eye to manipulate the outputs of a wide range of Natural Language Processing (NLP) systems from neural machine-translation pipelines to web search engines. We find that with a single imperceptible encoding injection -- representing one invisible character, homoglyph, reordering, or deletion -- an attacker can significantly reduce the performance of vulnerable models, and with three injections most models can be functionally broken. Our attacks work against currently-deployed commercial systems, including those produced by Microsoft and Google, in addition to open source models published by Facebook and IBM. This novel series of attacks presents a significant threat to many language processing systems: an attacker can affect systems in a targeted manner without any assumptions about the underlying model. We conclude that text-based NLP systems require careful input sanitization, just like conventional applications, and that given such systems are now being deployed rapidly at scale, the urgent attention of architects and operators is required.

(2 min) We propose an embarrassingly simple but very effective scheme for high-quality dense stereo reconstruction: (i) generate an approximate reconstruction with your favourite stereo matcher; (ii) rewarp the input images with that approximate model; (iii) with the initial reconstruction and the warped images as input, train a deep network to enhance the reconstruction by regressing a residual correction; and (iv) if desired, iterate the refinement with the new, improved reconstruction. The strategy to only learn the residual greatly simplifies the learning problem. A standard Unet without bells and whistles is enough to reconstruct even small surface details, like dormers and roof substructures in satellite images. We also investigate residual reconstruction with less information and find that even a single image is enough to greatly improve an approximate reconstruction. Our full model reduces the mean absolute error of state-of-the-art stereo reconstruction systems by >50%, both in our target domain of satellite stereo and on stereo pairs from the ETH3D benchmark.

(2 min) Despite the success of Generative Adversarial Networks (GANs), their training suffers from several well-known problems, including mode collapse and difficulties learning a disconnected set of manifolds. In this paper, we break down the challenging task of learning complex high dimensional distributions, supporting diverse data samples, to simpler sub-tasks. Our solution relies on designing a partitioner that breaks the space into smaller regions, each having a simpler distribution, and training a different generator for each partition. This is done in an unsupervised manner without requiring any labels. We formulate two desired criteria for the space partitioner that aid the training of our mixture of generators: 1) to produce connected partitions and 2) provide a proxy of distance between partitions and data samples, along with a direction for reducing that distance. These criteria are developed to avoid producing samples from places with non-existent data density, and also facilitate training by providing additional direction to the generators. We develop theoretical constraints for a space partitioner to satisfy the above criteria. Guided by our theoretical analysis, we design an effective neural architecture for the space partitioner that empirically assures these conditions. Experimental results on various standard benchmarks show that the proposed unsupervised model outperforms several recent methods.

(2 min) Recently, Vision Transformers (ViTs) have achieved impressive results on various vision tasks. Yet, their generalization ability under different distribution shifts is rarely understood. In this work, we provide a comprehensive study on the out-of-distribution generalization of ViTs. To support a systematic investigation, we first present a taxonomy of distribution shifts by categorizing them into five conceptual groups: corruption shift, background shift, texture shift, destruction shift, and style shift. Then we perform extensive evaluations of ViT variants under different groups of distribution shifts and compare their generalization ability with CNNs. Several important observations are obtained: 1) ViTs generalize better than CNNs under multiple distribution shifts. With the same or fewer parameters, ViTs are ahead of corresponding CNNs by more than 5% in top-1 accuracy under most distribution shifts. 2) Larger ViTs gradually narrow the in-distribution and out-of-distribution performance gap. To further improve the generalization of ViTs, we design the Generalization-Enhanced ViTs by integrating adversarial learning, information theory, and self-supervised learning. By investigating three types of generalization-enhanced ViTs, we observe their gradient-sensitivity and design a smoother learning strategy to achieve a stable training process. With modified training schemes, we achieve improvements on performance towards out-of-distribution data by 4% from vanilla ViTs. We comprehensively compare three generalization-enhanced ViTs with their corresponding CNNs, and observe that: 1) For the enhanced model, larger ViTs still benefit more for the out-of-distribution generalization. 2) generalization-enhanced ViTs are more sensitive to the hyper-parameters than corresponding CNNs. We hope our comprehensive study could shed light on the design of more generalizable learning architectures.

(2 min) We present a plug-in replacement for batch normalization (BN) called exponential moving average normalization (EMAN), which improves the performance of existing student-teacher based self- and semi-supervised learning techniques. Unlike the standard BN, where the statistics are computed within each batch, EMAN, used in the teacher, updates its statistics by exponential moving average from the BN statistics of the student. This design reduces the intrinsic cross-sample dependency of BN and enhances the generalization of the teacher. EMAN improves strong baselines for self-supervised learning by 4-6/1-2 points and semi-supervised learning by about 7/2 points, when 1%/10% supervised labels are available on ImageNet. These improvements are consistent across methods, network architectures, training duration, and datasets, demonstrating the general effectiveness of this technique. The code is available at https://github.com/amazon-research/exponential-moving-average-normalization.

(2 min) Place recognition is critical for both offline mapping and online localization. However, current single-sensor based place recognition still remains challenging in adverse conditions. In this paper, a heterogeneous measurements based framework is proposed for long-term place recognition, which retrieves the query radar scans from the existing lidar maps. To achieve this, a deep neural network is built with joint training in the learning stage, and then in the testing stage, shared embeddings of radar and lidar are extracted for heterogeneous place recognition. To validate the effectiveness of the proposed method, we conduct tests and generalization experiments on the multi-session public datasets compared to other competitive methods. The experimental results indicate that our model is able to perform multiple place recognitions: lidar-to-lidar, radar-to-radar and radar-to-lidar, while the learned model is trained only once. We also release the source code publicly: https://github.com/ZJUYH/radar-to-lidar-place-recognition.

(2 min) Collecting a large number of reliable training images annotated by multiple land-cover class labels in the framework of multi-label classification is time-consuming and costly in remote sensing (RS). To address this problem, publicly available thematic products are often used for annotating RS images with zero-labeling-cost. However, such an approach may result in constructing a training set with noisy multi-labels, distorting the learning process. To address this problem, we propose a Consensual Collaborative Multi-Label Learning (CCML) method. The proposed CCML identifies, ranks and corrects training images with noisy multi-labels through four main modules: 1) discrepancy module; 2) group lasso module; 3) flipping module; and 4) swap module. The discrepancy module ensures that the two networks learn diverse features, while obtaining the same predictions. The group lasso module detects the potentially noisy labels by estimating the label uncertainty based on the aggregation of two collaborative networks. The flipping module corrects the identified noisy labels, whereas the swap module exchanges the ranking information between the two networks. The experimental results confirm the success of the proposed CCML under high (synthetically added) multi-label noise rates. The code of the proposed method is publicly available at https://noisy-labels-in-rs.org

(2 min) In adversarial training (AT), the main focus has been the objective and optimizer while the model has been less studied, so that the models being used are still those classic ones in standard training (ST). Classic network architectures (NAs) are generally worse than searched NAs in ST, which should be the same in AT. In this paper, we argue that NA and AT cannot be handled independently, since given a dataset, the optimal NA in ST would be no longer optimal in AT. That being said, AT is time-consuming itself; if we directly search NAs in AT over large search spaces, the computation will be practically infeasible. Thus, we propose a diverse-structured network (DS-Net), to significantly reduce the size of the search space: instead of low-level operations, we only consider predefined atomic blocks, where an atomic block is a time-tested building block like the residual block. There are only a few atomic blocks and thus we can weight all atomic blocks rather than find the best one in a searched block of DS-Net, which is an essential trade-off between exploring diverse structures and exploiting the best structures. Empirical results demonstrate the advantages of DS-Net, i.e., weighting the atomic blocks.

(2 min) Despite the great progress made by deep neural networks in the semantic segmentation task, traditional neural-networkbased methods typically suffer from a shortage of large amounts of pixel-level annotations. Recent progress in fewshot semantic segmentation tackles the issue by only a few pixel-level annotated examples. However, these few-shot approaches cannot easily be applied to multi-way or weak annotation settings. In this paper, we advance the few-shot segmentation paradigm towards a scenario where image-level annotations are available to help the training process of a few pixel-level annotations. Our key idea is to learn a better prototype representation of the class by fusing the knowledge from the image-level labeled data. Specifically, we propose a new framework, called PAIA, to learn the class prototype representation in a metric space by integrating image-level annotations. Furthermore, by considering the uncertainty of pseudo-masks, a distilled soft masked average pooling strategy is designed to handle distractions in image-level annotations. Extensive empirical results on two datasets show superior performance of PAIA.

(2 min) This work explores facial expression bias as a security vulnerability of face recognition systems. Despite the great performance achieved by state-of-the-art face recognition systems, the algorithms are still sensitive to a large range of covariates. We present a comprehensive analysis of how facial expression bias impacts the performance of face recognition technologies. Our study analyzes: i) facial expression biases in the most popular face recognition databases; and ii) the impact of facial expression in face recognition performances. Our experimental framework includes two face detectors, three face recognition models, and three different databases. Our results demonstrate a huge facial expression bias in the most widely used databases, as well as a related impact of face expression in the performance of state-of-the-art algorithms. This work opens the door to new research lines focused on mitigating the observed vulnerability.

(2 min) Retinal artery/vein (A/V) classification is a critical technique for diagnosing diabetes and cardiovascular diseases. Although deep learning based methods achieve impressive results in A/V classification, their performances usually degrade severely when being directly applied to another database, due to the domain shift, e.g., caused by the variations in imaging protocols. In this paper, we propose a novel vessel-mixing based consistency regularization framework, for cross-domain learning in retinal A/V classification. Specially, to alleviate the severe bias to source domain, based on the label smooth prior, the model is regularized to give consistent predictions for unlabeled target-domain inputs that are under perturbation. This consistency regularization implicitly introduces a mechanism where the model and the perturbation is opponent to each other, where the model is pushed to be robust enough to cope with the perturbation. Thus, we investigate a more difficult opponent to further inspire the robustness of model, in the scenario of retinal A/V, called vessel-mixing perturbation. Specially, it effectively disturbs the fundus images especially the vessel structures by mixing two images regionally. We conduct extensive experiments on cross-domain A/V classification using four public datasets, which are collected by diverse institutions and imaging devices. The results demonstrate that our method achieves the state-of-the-art cross-domain performance, which is also close to the upper bound obtained by fully supervised learning on target domain.

(3 min) The determination of precise skin lesion boundaries in dermoscopic images using automated methods faces many challenges, most importantly, the presence of hair, inconspicuous lesion edges and low contrast in dermoscopic images, and variability in the color, texture and shapes of skin lesions. Existing deep learning-based skin lesion segmentation algorithms are expensive in terms of computational time and memory. Consequently, running such segmentation algorithms requires a powerful GPU and high bandwidth memory, which are not available in dermoscopy devices. Thus, this article aims to achieve precise skin lesion segmentation with minimum resources: a lightweight, efficient generative adversarial network (GAN) model called SLSNet, which combines 1-D kernel factorized networks, position and channel attention, and multiscale aggregation mechanisms with a GAN model. The 1-D kernel factorized network reduces the computational cost of 2D filtering. The position and channel attention modules enhance the discriminative ability between the lesion and non-lesion feature representations in spatial and channel dimensions, respectively. A multiscale block is also used to aggregate the coarse-to-fine features of input skin images and reduce the effect of the artifacts. SLSNet is evaluated on two publicly available datasets: ISBI 2017 and the ISIC 2018. Although SLSNet has only 2.35 million parameters, the experimental results demonstrate that it achieves segmentation results on a par with the state-of-the-art skin lesion segmentation methods with an accuracy of 97.61%, and Dice and Jaccard similarity coefficients of 90.63% and 81.98%, respectively. SLSNet can run at more than 110 frames per second (FPS) in a single GTX1080Ti GPU, which is faster than well-known deep learning-based image segmentation models, such as FCN. Therefore, SLSNet can be used for practical dermoscopic applications.

(2 min) Localizing stereo boundaries is difficult because matching cues are absent in the occluded regions that are adjacent to them. We introduce an energy and level-set optimizer that improves boundaries by encoding the essential geometry of occlusions: The spatial extent of an occlusion must equal the amplitude of the disparity jump that causes it. In a collection of figure-ground scenes from Middlebury and Falling Things stereo datasets, the model provides more accurate boundaries than previous occlusion-handling techniques.

(2 min) We define the concept of CompositeTasking as the fusion of multiple, spatially distributed tasks, for various aspects of image understanding. Learning to perform spatially distributed tasks is motivated by the frequent availability of only sparse labels across tasks, and the desire for a compact multi-tasking network. To facilitate CompositeTasking, we introduce a novel task conditioning model -- a single encoder-decoder network that performs multiple, spatially varying tasks at once. The proposed network takes an image and a set of pixel-wise dense task requests as inputs, and performs the requested prediction task for each pixel. Moreover, we also learn the composition of tasks that needs to be performed according to some CompositeTasking rules, which includes the decision of where to apply which task. It not only offers us a compact network for multi-tasking, but also allows for task-editing. Another strength of the proposed method is demonstrated by only having to supply sparse supervision per task. The obtained results are on par with our baselines that use dense supervision and a multi-headed multi-tasking design. The source code will be made publicly available at www.github.com/nikola3794/composite-tasking.

(2 min) Photoacoustic imaging has the potential to revolutionise healthcare due to the valuable information on tissue physiology that is contained in multispectral photoacoustic measurements. Clinical translation of the technology requires conversion of the high-dimensional acquired data into clinically relevant and interpretable information. In this work, we present a deep learning-based approach to semantic segmentation of multispectral photoacoustic images to facilitate the interpretability of recorded images. Manually annotated multispectral photoacoustic imaging data are used as gold standard reference annotations and enable the training of a deep learning-based segmentation algorithm in a supervised manner. Based on a validation study with experimentally acquired data of healthy human volunteers, we show that automatic tissue segmentation can be used to create powerful analyses and visualisations of multispectral photoacoustic images. Due to the intuitive representation of high-dimensional information, such a processing algorithm could be a valuable means to facilitate the clinical translation of photoacoustic imaging.

(2 min) Radars and cameras are mature, cost-effective, and robust sensors and have been widely used in the perception stack of mass-produced autonomous driving systems. Due to their complementary properties, outputs from radar detection (radar pins) and camera perception (2D bounding boxes) are usually fused to generate the best perception results. The key to successful radar-camera fusion is the accurate data association. The challenges in the radar-camera association can be attributed to the complexity of driving scenes, the noisy and sparse nature of radar measurements, and the depth ambiguity from 2D bounding boxes. Traditional rule-based association methods are susceptible to performance degradation in challenging scenarios and failure in corner cases. In this study, we propose to address radar-camera association via deep representation learning, to explore feature-level interaction and global reasoning. Additionally, we design a loss sampling mechanism and an innovative ordinal loss to overcome the difficulty of imperfect labeling and to enforce critical human-like reasoning. Despite being trained with noisy labels generated by a rule-based algorithm, our proposed method achieves a performance of 92.2% F1 score, which is 11.6% higher than the rule-based teacher. Moreover, this data-driven method also lends itself to continuous improvement via corner case mining.

(2 min) 3D Point cloud registration is still a very challenging topic due to the difficulty in finding the rigid transformation between two point clouds with partial correspondences, and it's even harder in the absence of any initial estimation information. In this paper, we present an end-to-end deep-learning based approach to resolve the point cloud registration problem. Firstly, the revised LPD-Net is introduced to extract features and aggregate them with the graph network. Secondly, the self-attention mechanism is utilized to enhance the structure information in the point cloud and the cross-attention mechanism is designed to enhance the corresponding information between the two input point clouds. Based on which, the virtual corresponding points can be generated by a soft pointer based method, and finally, the point cloud registration problem can be solved by implementing the SVD method. Comparison results in ModelNet40 dataset validate that the proposed approach reaches the state-of-the-art in point cloud registration tasks and experiment resutls in KITTI dataset validate the effectiveness of the proposed approach in real applications.Our source code is available at \url{https://github.com/qiaozhijian/VCR-Net.git}

(2 min) We present VSAC, a RANSAC-type robust estimator with a number of novelties. It benefits from the introduction of the concept of independent inliers that improves significantly the efficacy of the dominant plane handling and, also, allows near error-free rejection of incorrect models, without false positives. The local optimization process and its application is improved so that it is run on average only once. Further technical improvements include adaptive sequential hypothesis verification and efficient model estimation via Gaussian elimination. Experiments on four standard datasets show that VSAC is significantly faster than all its predecessors and runs on average in 1-2 ms, on a CPU. It is two orders of magnitude faster and yet as precise as MAGSAC++, the currently most accurate estimator of two-view geometry. In the repeated runs on EVD, HPatches, PhotoTourism, and Kusvod2 datasets, it never failed.

(2 min) We propose associating language utterances to 3D visual abstractions of the scene they describe. The 3D visual abstractions are encoded as 3-dimensional visual feature maps. We infer these 3D visual scene feature maps from RGB images of the scene via view prediction: when the generated 3D scene feature map is neurally projected from a camera viewpoint, it should match the corresponding RGB image. We present generative models that condition on the dependency tree of an utterance and generate a corresponding visual 3D feature map as well as reason about its plausibility, and detector models that condition on both the dependency tree of an utterance and a related image and localize the object referents in the 3D feature map inferred from the image. Our model outperforms models of language and vision that associate language with 2D CNN activations or 2D images by a large margin in a variety of tasks, such as, classifying plausibility of utterances, detecting referential expressions, and supplying rewards for trajectory optimization of object placement policies from language instructions. We perform numerous ablations and show the improved performance of our detectors is due to its better generalization across camera viewpoints and lack of object interferences in the inferred 3D feature space, and the improved performance of our generators is due to their ability to spatially reason about objects and their configurations in 3D when mapping from language to scenes.

(2 min) Recently normalizing flows (NFs) have demonstrated state-of-the-art performance on modeling 3D point clouds while allowing sampling with arbitrary resolution at inference time. However, these flow-based models still require long training times and large models for representing complicated geometries. This work enhances their representational power by applying mixtures of NFs to point clouds. We show that in this more general framework each component learns to specialize in a particular subregion of an object in a completely unsupervised fashion. By instantiating each mixture component with a comparatively small NF we generate point clouds with improved details compared to single-flow-based models while using fewer parameters and considerably reducing the inference runtime. We further demonstrate that by adding data augmentation, individual mixture components can learn to specialize in a semantically meaningful manner. We evaluate mixtures of NFs on generation, autoencoding and single-view reconstruction based on the ShapeNet dataset.

(2 min) Recent work in equivariant deep learning bears strong similarities to physics. Fields over a base space are fundamental entities in both subjects, as are equivariant maps between these fields. In deep learning, however, these maps are usually defined by convolutions with a kernel, whereas they are partial differential operators (PDOs) in physics. Developing the theory of equivariant PDOs in the context of deep learning could bring these subjects even closer together and lead to a stronger flow of ideas. In this work, we derive a $G$-steerability constraint that completely characterizes when a PDO between feature vector fields is equivariant, for arbitrary symmetry groups $G$. We then fully solve this constraint for several important groups. We use our solutions as equivariant drop-in replacements for convolutional layers and benchmark them in that role. Finally, we develop a framework for equivariant maps based on Schwartz distributions that unifies classical convolutions and differential operators and gives insight about the relation between the two.

(2 min) Longitudinal MRIs are often used to capture the gradual deterioration of brain structure and function caused by aging or neurological diseases. Analyzing this data via machine learning generally requires a large number of ground-truth labels, which are often missing or expensive to obtain. Reducing the need for labels, we propose a self-supervised strategy for representation learning named Longitudinal Neighborhood Embedding (LNE). Motivated by concepts in contrastive learning, LNE explicitly models the similarity between trajectory vectors across different subjects. We do so by building a graph in each training iteration defining neighborhoods in the latent space so that the progression direction of a subject follows the direction of its neighbors. This results in a smooth trajectory field that captures the global morphological change of the brain while maintaining the local continuity. We apply LNE to longitudinal T1w MRIs of two neuroimaging studies: a dataset composed of 274 healthy subjects, and Alzheimer's Disease Neuroimaging Initiative (ADNI, N=632). The visualization of the smooth trajectory vector field and superior performance on downstream tasks demonstrate the strength of the proposed method over existing self-supervised methods in extracting information associated with normal aging and in revealing the impact of neurodegenerative disorders. The code is available at \url{https://github.com/ouyangjiahong/longitudinal-neighbourhood-embedding.git}.

(2 min) Deep learning has demonstrated its power in image rectification by leveraging the representation capacity of deep neural networks via supervised training based on a large-scale synthetic dataset. However, the model may overfit the synthetic images and generalize not well on real-world fisheye images due to the limited universality of a specific distortion model and the lack of explicitly modeling the distortion and rectification process. In this paper, we propose a novel self-supervised image rectification (SIR) method based on an important insight that the rectified results of distorted images of a same scene from different lens should be the same. Specifically, we devise a new network architecture with a shared encoder and several prediction heads, each of which predicts the distortion parameter of a specific distortion model. We further leverage a differentiable warping module to generate the rectified images and re-distorted images from the distortion parameters and exploit the intra- and inter-model consistency between them during training, thereby leading to a self-supervised learning scheme without the need for ground-truth distortion parameters or normal images. Experiments on synthetic dataset and real-world fisheye images demonstrate that our method achieves comparable or even better performance than the supervised baseline method and representative state-of-the-art methods. Self-supervised learning also improves the universality of distortion models while keeping their self-consistency.

(2 min) Multi-source unsupervised domain adaptation~(MSDA) aims at adapting models trained on multiple labeled source domains to an unlabeled target domain. In this paper, we propose a novel multi-source domain adaptation framework based on collaborative learning for semantic segmentation. Firstly, a simple image translation method is introduced to align the pixel value distribution to reduce the gap between source domains and target domain to some extent. Then, to fully exploit the essential semantic information across source domains, we propose a collaborative learning method for domain adaptation without seeing any data from target domain. In addition, similar to the setting of unsupervised domain adaptation, unlabeled target domain data is leveraged to further improve the performance of domain adaptation. This is achieved by additionally constraining the outputs of multiple adaptation models with pseudo labels online generated by an ensembled model. Extensive experiments and ablation studies are conducted on the widely-used domain adaptation benchmark datasets in semantic segmentation. Our proposed method achieves 59.0\% mIoU on the validation set of Cityscapes by training on the labeled Synscapes and GTA5 datasets and unlabeled training set of Cityscapes. It significantly outperforms all previous state-of-the-arts single-source and multi-source unsupervised domain adaptation methods.

(2 min) Despite its best performance in image denoising, the supervised deep denoising methods require paired noise-clean data, which are often unavailable. To address this challenge, Noise2Noise was designed based on the fact that paired noise-clean images can be replaced by paired noise-noise images that are easier to collect. However, in many scenarios the collection of paired noise-noise images is still impractical. To bypass labeled images, Noise2Void methods predict masked pixels from their surroundings with single noisy images only and give improved denoising results that still need improvements. An observation on classic denoising methods is that non-local mean (NLM) outcomes are typically superior to locally denoised results. In contrast, Noise2Void and its variants do not utilize self-similarities in an image as the NLM-based methods do. Here we propose Noise2Sim, an NLM-inspired self-learning method for image denoising. Specifically, Noise2Sim leverages the self-similarity of image pixels to train the denoising network, requiring single noisy images only. Our theoretical analysis shows that Noise2Sim tends to be equivalent to Noise2Noise under mild conditions. To efficiently manage the computational burden for globally searching similar pixels, we design a two-step procedure to provide data for Noise2Sim training. Extensive experiments demonstrate the superiority of Noise2Sim on common benchmark datasets.

(2 min) Pedestrian trajectory prediction in dynamic scenes remains a challenging and critical problem in numerous applications, such as self-driving cars and socially aware robots. Challenges concentrate on capturing pedestrians' motion patterns and social interactions, as well as handling the future uncertainties. Recent studies focus on modeling pedestrians' motion patterns with recurrent neural networks, capturing social interactions with pooling-based or graph-based methods, and handling future uncertainties by using random Gaussian noise as the latent variable. However, they do not integrate specific obstacle avoidance experience (OAE) that may improve prediction performance. For example, pedestrians' future trajectories are always influenced by others in front. Here we propose GTPPO (Graph-based Trajectory Predictor with Pseudo Oracle), an encoder-decoder-based method conditioned on pedestrians' future behaviors. Pedestrians' motion patterns are encoded with a long short-term memory unit, which introduces the temporal attention to highlight specific time steps. Their interactions are captured by a graph-based attention mechanism, which draws OAE into the data-driven learning process of graph attention. Future uncertainties are handled by generating multi-modal outputs with an informative latent variable. Such a variable is generated by a novel pseudo oracle predictor, which minimizes the knowledge gap between historical and ground-truth trajectories. Finally, the GTPPO is evaluated on ETH, UCY and Stanford Drone datasets, and the results demonstrate state-of-the-art performance. Besides, the qualitative evaluations show successful cases of handling sudden motion changes in the future. Such findings indicate that GTPPO can peek into the future.

(2 min) Traffic near-crash events serve as critical data sources for various smart transportation applications, such as being surrogate safety measures for traffic safety research and corner case data for automated vehicle testing. However, there are several key challenges for near-crash detection. First, extracting near-crashes from original data sources requires significant computing, communication, and storage resources. Also, existing methods lack efficiency and transferability, which bottlenecks prospective large-scale applications. To this end, this paper leverages the power of edge computing to address these challenges by processing the video streams from existing dashcams onboard in a real-time manner. We design a multi-thread system architecture that operates on edge devices and model the bounding boxes generated by object detection and tracking in linear complexity. The method is insensitive to camera parameters and backward compatible with different vehicles. The edge computing system has been evaluated with recorded videos and real-world tests on two cars and four buses for over ten thousand hours. It filters out irrelevant videos in real-time thereby saving labor cost, processing time, network bandwidth, and data storage. It collects not only event videos but also other valuable data such as road user type, event location, time to collision, vehicle trajectory, vehicle speed, brake switch, and throttle. The experiments demonstrate the promising performance of the system regarding efficiency, accuracy, reliability, and transferability. It is among the first efforts in applying edge computing for real-time traffic video analytics and is expected to benefit multiple sub-fields in smart transportation research and applications.

(2 min) Recent work has shown that deep learning models can be used to classify land-use data from geospatial satellite imagery. We show that when these deep learning models are trained on data from specific continents/seasons, there is a high degree of variability in model performance on out-of-sample continents/seasons. This suggests that just because a model accurately predicts land-use classes in one continent or season does not mean that the model will accurately predict land-use classes in a different continent or season. We then use clustering techniques on satellite imagery from different continents to visualize the differences in landscapes that make geospatial generalization particularly difficult, and summarize our takeaways for future satellite imagery-related applications.

(2 min) Vision Transformers (ViT) have been shown to attain highly competitive performance for a wide range of vision applications, such as image classification, object detection and semantic image segmentation. In comparison to convolutional neural networks, the Vision Transformer's weaker inductive bias is generally found to cause an increased reliance on model regularization or data augmentation (``AugReg'' for short) when training on smaller training datasets. We conduct a systematic empirical study in order to better understand the interplay between the amount of training data, AugReg, model size and compute budget. As one result of this study we find that the combination of increased compute and AugReg can yield models with the same performance as models trained on an order of magnitude more training data: we train ViT models of various sizes on the public ImageNet-21k dataset which either match or outperform their counterparts trained on the larger, but not publicly available JFT-300M dataset.

(2 min) Nearest neighbor search is to find the data points in the database such that the distances from them to the query are the smallest, which is a fundamental problem in various domains, such as computer vision, recommendation systems and machine learning. Hashing is one of the most widely used methods for its computational and storage efficiency. With the development of deep learning, deep hashing methods show more advantages than traditional methods. In this paper, we present a comprehensive survey of the deep hashing algorithms. Specifically, we categorize deep supervised hashing methods into pairwise similarity preserving, multiwise similarity preserving, implicit similarity preserving, classification-oriented preserving as well as quantization according to the manners of preserving the similarities. In addition, we also introduce some other topics such as deep unsupervised hashing and multi-modal deep hashing methods. Meanwhile, we also present some commonly used public datasets and the scheme to measure the performance of deep hashing algorithms. Finally, we discussed some potential research directions in conclusion.

(2 min) While medical image segmentation is an important task for computer aided diagnosis, the high expertise requirement for pixelwise manual annotations makes it a challenging and time consuming task. Since conventional data augmentations do not fully represent the underlying distribution of the training set, the trained models have varying performance when tested on images captured from different sources. Most prior work on image synthesis for data augmentation ignore the interleaved geometric relationship between different anatomical labels. We propose improvements over previous GAN-based medical image synthesis methods by learning the relationship between different anatomical labels. We use a weakly supervised segmentation method to obtain pixel level semantic label map of images which is used learn the intrinsic relationship of geometry and shape across semantic labels. Latent space variable sampling results in diverse generated images from a base image and improves robustness. We use the synthetic images from our method to train networks for segmenting COVID-19 infected areas from lung CT images. The proposed method outperforms state-of-the-art segmentation methods on a public dataset. Ablation studies also demonstrate benefits of integrating geometry and diversity.

(2 min) Pseudo-healthy synthesis is the task of creating a subject-specific `healthy' image from a pathological one. Such images can be helpful in tasks such as anomaly detection and understanding changes induced by pathology and disease. In this paper, we present a model that is encouraged to disentangle the information of pathology from what seems to be healthy. We disentangle what appears to be healthy and where disease is as a segmentation map, which are then recombined by a network to reconstruct the input disease image. We train our models adversarially using either paired or unpaired settings, where we pair disease images and maps when available. We quantitatively and subjectively, with a human study, evaluate the quality of pseudo-healthy images using several criteria. We show in a series of experiments, performed on ISLES, BraTS and Cam-CAN datasets, that our method is better than several baselines and methods from the literature. We also show that due to better training processes we could recover deformations, on surrounding tissue, caused by disease. Our implementation is publicly available at https://github.com/xiat0616/pseudo-healthy-synthesis. This paper has been accepted by Medical Image Analysis: https://doi.org/10.1016/j.media.2020.101719.

(2 min) Most capsule network designs rely on traditional matrix multiplication between capsule layers and computationally expensive routing mechanisms to deal with the capsule dimensional entanglement that the matrix multiplication introduces. By using Homogeneous Vector Capsules (HVCs), which use element-wise multiplication rather than matrix multiplication, the dimensions of the capsules remain unentangled. In this work, we study HVCs as applied to the highly structured MNIST dataset in order to produce a direct comparison to the capsule research direction of Geoffrey Hinton, et al. In our study, we show that a simple convolutional neural network using HVCs performs as well as the prior best performing capsule network on MNIST using 5.5x fewer parameters, 4x fewer training epochs, no reconstruction sub-network, and requiring no routing mechanism. The addition of multiple classification branches to the network establishes a new state of the art for the MNIST dataset with an accuracy of 99.87% for an ensemble of these models, as well as establishing a new state of the art for a single model (99.83% accurate).

(2 min) With the rise of deep learning models in the field of computer vision, new possibilities for their application in industrial processes proves to return great benefits. Nevertheless, the actual fit of machine learning for highly standardised industrial processes is still under debate. This paper addresses the challenges on the industrial realization of the AI tools, considering the use case of Laser Beam Welding quality control as an example. We use object detection algorithms from the TensorFlow object detection API and adapt them to our use case using transfer learning. The baseline models we develop are used as benchmarks and evaluated and compared to models that undergo dataset scaling and hyperparameter tuning. We find that moderate scaling of the dataset via image augmentation leads to improvements in intersection over union (IoU) and recall, whereas high levels of augmentation and scaling may lead to deterioration of results. Finally, we put our results into perspective of the underlying use case and evaluate their fit.

(2 min) Temporal action detection (TAD) aims to determine the semantic label and the boundaries of every action instance in an untrimmed video. It is a fundamental task in video understanding and significant progress has been made in TAD. Previous methods involve multiple stages or networks and hand-designed rules or operations, which fall short in efficiency and flexibility. Here, we construct an end-to-end framework for TAD upon Transformer, termed \textit{TadTR}, which simultaneously predicts all action instances as a set of labels and temporal locations in parallel. TadTR is able to adaptively extract temporal context information needed for making action predictions, by selectively attending to a number of snippets in a video. It greatly simplifies the pipeline of TAD and runs much faster than previous detectors. Our method achieves state-of-the-art performance on HACS Segments and THUMOS14 and competitive performance on ActivityNet-1.3. Our code will be made available at \url{https://github.com/xlliu7/TadTR}.

(2 min) Instance-level contrastive learning techniques, which rely on data augmentation and a contrastive loss function, have found great success in the domain of visual representation learning. They are not suitable for exploiting the rich dynamical structure of video however, as operations are done on many augmented instances. In this paper we propose "Video Cross-Stream Prototypical Contrasting", a novel method which predicts consistent prototype assignments from both RGB and optical flow views, operating on sets of samples. Specifically, we alternate the optimization process; while optimizing one of the streams, all views are mapped to one set of stream prototype vectors. Each of the assignments is predicted with all views except the one matching the prediction, pushing representations closer to their assigned prototypes. As a result, more efficient video embeddings with ingrained motion information are learned, without the explicit need for optical flow computation during inference. We obtain state-of-the-art results on nearest neighbour video retrieval and action recognition, outperforming previous best by +3.2% on UCF101 using the S3D backbone (90.5% Top-1 acc), and by +7.2% on UCF101 and +15.1% on HMDB51 using the R(2+1)D backbone.

(2 min) The breakthrough of contrastive learning (CL) has fueled the recent success of self-supervised learning (SSL) in high-level vision tasks on RGB images. However, CL is still ill-defined for low-level vision tasks, such as joint demosaicking and denoising (JDD), in the RAW domain. To bridge this methodological gap, we present a novel CL approach on RAW images, residual contrastive learning (RCL), which aims to learn meaningful representations for JDD. Our work is built on the assumption that noise contained in each RAW image is signal-dependent, thus two crops from the same RAW image should have more similar noise distribution than two crops from different RAW images. We use residuals as a discriminative feature and the earth mover's distance to measure the distribution divergence for the contrastive loss. To evaluate the proposed CL strategy, we simulate a series of unsupervised JDD experiments with large-scale data corrupted by synthetic signal-dependent noise, where we set a new benchmark for unsupervised JDD tasks with unknown (random) noise variance. Our empirical study not only validates that CL can be applied on distributions (c.f. features), but also exposes the lack of robustness of previous non-ML and SSL JDD methods when the statistics of the noise are unknown, thus providing some further insight into signal-dependent noise problems.

(2 min) The demand for on-device document recognition systems increases in conjunction with the emergence of more strict privacy and security requirements. In such systems, there is no data transfer from the end device to a third-party information processing servers. The response time is vital to the user experience of on-device document recognition. Combined with the unavailability of discrete GPUs, powerful CPUs, or a large RAM capacity on consumer-grade end devices such as smartphones, the time limitations put significant constraints on the computational complexity of the applied algorithms for on-device execution. In this work, we consider document location in an image without prior knowledge of the document content or its internal structure. In accordance with the published works, at least 5 systems offer solutions for on-device document location. All these systems use a location method which can be considered Hough-based. The precision of such systems seems to be lower than that of the state-of-the-art solutions which were not designed to account for the limited computational resources. We propose an advanced Hough-based method. In contrast with other approaches, it accounts for the geometric invariants of the central projection model and combines both edge and color features for document boundary detection. The proposed method allowed for the second best result for SmartDoc dataset in terms of precision, surpassed by U-net like neural network. When evaluated on a more challenging MIDV-500 dataset, the proposed algorithm guaranteed the best precision compared to published methods. Our method retained the applicability to on-device computations.

(2 min) Boundary-based instance segmentation has drawn much attention since of its attractive efficiency. However, existing methods suffer from the difficulty in long-distance regression. In this paper, we propose a coarse-to-fine module to address the problem. Approximate boundary points are generated at the coarse stage and then features of these points are sampled and fed to a refined regressor for fine prediction. It is end-to-end trainable since differential sampling operation is well supported in the module. Furthermore, we design a holistic boundary-aware branch and introduce instance-agnostic supervision to assist regression. Equipped with ResNet-101, our approach achieves 31.7\% mask AP on COCO dataset with single-scale training and testing, outperforming the baseline 1.3\% mask AP with less than 1\% additional parameters and GFLOPs. Experiments also show that our proposed method achieves competitive performance compared to existing boundary-based methods with a lightweight design and a simple pipeline.

(2 min) Combining Natural Language with Vision represents a unique and interesting challenge in the domain of Artificial Intelligence. The AI City Challenge Track 5 for Natural Language-Based Vehicle Retrieval focuses on the problem of combining visual and textual information, applied to a smart-city use case. In this paper, we present All You Can Embed (AYCE), a modular solution to correlate single-vehicle tracking sequences with natural language. The main building blocks of the proposed architecture are (i) BERT to provide an embedding of the textual descriptions, (ii) a convolutional backbone along with a Transformer model to embed the visual information. For the training of the retrieval model, a variation of the Triplet Margin Loss is proposed to learn a distance measure between the visual and language embeddings. The code is publicly available at https://github.com/cscribano/AYCE_2021.

(2 min) Representing games through their pixels offers a promising approach for building general-purpose and versatile game models. While games are not merely images, neural network models trained on game pixels often capture differences of the visual style of the image rather than the content of the game. As a result, such models cannot generalize well even within similar games of the same genre. In this paper we build on recent advances in contrastive learning and showcase its benefits for representation learning in games. Learning to contrast images of games not only classifies games in a more efficient manner; it also yields models that separate games in a more meaningful fashion by ignoring the visual style and focusing, instead, on their content. Our results in a large dataset of sports video games containing 100k images across 175 games and 10 game genres suggest that contrastive learning is better suited for learning generalized game representations compared to conventional supervised learning. The findings of this study bring us closer to universal visual encoders for games that can be reused across previously unseen games without requiring retraining or fine-tuning.

(2 min) When interacting with objects through cameras, or pictures, users often have a specific intent. For example, they may want to perform a visual search. However, most object detection models ignore the user intent, relying on image pixels as their only input. This often leads to incorrect results, such as lack of a high-confidence detection on the object of interest, or detection with a wrong class label. In this paper we investigate techniques to modulate standard object detectors to explicitly account for the user intent, expressed as an embedding of a simple query. Compared to standard object detectors, query-modulated detectors show superior performance at detecting objects for a given label of interest. Thanks to large-scale training data synthesized from standard object detection annotations, query-modulated detectors can also outperform specialized referring expression recognition systems. Furthermore, they can be simultaneously trained to solve for both query-modulated detection and standard object detection.

(2 min) Throughout the COVID-19 pandemic, the most common symptom displayed by patients has been a fever, leading to the use of temperature scanning as a preemptive measure to detect potential carriers of the virus. Human employees with handheld thermometers have been used to fulfill this task, however this puts them at risk as they cannot be physically distanced and the sequential nature of this method leads to great inconveniences and inefficiency. The proposed solution is an autonomously navigating robot capable of conversing and scanning people's temperature to detect fevers and help screen for COVID-19. To satisfy this objective, the robot must be able to (1) navigate autonomously, (2) detect and track people, and (3) get individuals' temperature reading and converse with them if it exceeds 38{\deg}C. An autonomously navigating mobile robot is used with a manipulator controlled using a face tracking algorithm, and an end effector consisting of a thermal camera, smartphone, and chatbot. The goal is to develop a functioning solution that performs the above tasks. In addition, technical challenges encountered and their engineering solutions will be presented, and recommendations will be made for enhancements that could be incorporated when approaching commercialization.

(2 min) Recently, autonomous vehicles and those equipped with an Advanced Driver Assistance System (ADAS) are emerging. They share the road with regular ones operated by human drivers entirely. To ensure guaranteed safety for passengers and other road users, it becomes essential for autonomous vehicles and ADAS to anticipate traffic accidents from natural driving scenes. The dynamic spatial-temporal interaction of the traffic agents is complex, and visual cues for predicting a future accident are embedded deeply in dashcam video data. Therefore, early anticipation of traffic accidents remains a challenge. To this end, the paper presents a dynamic spatial-temporal attention (DSTA) network for early anticipation of traffic accidents from dashcam videos. The proposed DSTA-network learns to select discriminative temporal segments of a video sequence with a module named Dynamic Temporal Attention (DTA). It also learns to focus on the informative spatial regions of frames with another module named Dynamic Spatial Attention (DSA). The spatial-temporal relational features of accidents, along with scene appearance features, are learned jointly with a Gated Recurrent Unit (GRU) network. The experimental evaluation of the DSTA-network on two benchmark datasets confirms that it has exceeded the state-of-the-art performance. A thorough ablation study evaluates the contributions of individual components of the DSTA-network, revealing how the network achieves such performance. Furthermore, this paper proposes a new strategy that fuses the prediction scores from two complementary models and verifies its effectiveness in further boosting the performance of early accident anticipation.

(2 min) Many applications require the robustness, or ideally the invariance, of a neural network to certain transformations of input data. Most commonly, this requirement is addressed by either augmenting the training data, using adversarial training, or defining network architectures that include the desired invariance automatically. Unfortunately, the latter often relies on the ability to enlist all possible transformations, which make such approaches largely infeasible for infinite sets of transformations, such as arbitrary rotations or scaling. In this work, we propose a method for provably invariant network architectures with respect to group actions by choosing one element from a (possibly continuous) orbit based on a fixed criterion. In a nutshell, we intend to 'undo' any possible transformation before feeding the data into the actual network. We analyze properties of such approaches, extend them to equivariant networks, and demonstrate their advantages in terms of robustness as well as computational efficiency in several numerical examples. In particular, we investigate the robustness with respect to rotations of images (which can possibly hold up to discretization artifacts only) as well as the provable rotational and scaling invariance of 3D point cloud classification.

(2 min) Nighttime photographers are often troubled by light pollution of unwanted artificial lights. Artificial lights, after scattered by aerosols in the atmosphere, can inundate the starlight and degrade the quality of nighttime images, by reducing contrast and dynamic range and causing hazes. In this paper we develop a physically-based light pollution reduction (LPR) algorithm that can substantially alleviate the aforementioned degradations of perceptual quality and restore the pristine state of night sky. The key to the success of the proposed LPR algorithm is an inverse method to estimate the spatial radiance distribution and spectral signature of ground artificial lights. Extensive experiments are carried out to evaluate the efficacy and limitations of the LPR algorithm.

(2 min) Despite the significant advances in deep learning over the past decade, a major challenge that limits the wide-spread adoption of deep learning has been their fragility to adversarial attacks. This sensitivity to making erroneous predictions in the presence of adversarially perturbed data makes deep neural networks difficult to adopt for certain real-world, mission-critical applications. While much of the research focus has revolved around adversarial example creation and adversarial hardening, the area of performance measures for assessing adversarial robustness is not well explored. Motivated by this, this study presents the concept of residual error, a new performance measure for not only assessing the adversarial robustness of a deep neural network at the individual sample level, but also can be used to differentiate between adversarial and non-adversarial examples to facilitate for adversarial example detection. Furthermore, we introduce a hybrid model for approximating the residual error in a tractable manner. Experimental results using the case of image classification demonstrates the effectiveness and efficacy of the proposed residual error metric for assessing several well-known deep neural network architectures. These results thus illustrate that the proposed measure could be a useful tool for not only assessing the robustness of deep neural networks used in mission-critical scenarios, but also in the design of adversarially robust models.

(2 min) Reconstruction of object or scene surfaces has tremendous applications in computer vision, computer graphics, and robotics. In this paper, we study a fundamental problem in this context about recovering a surface mesh from an implicit field function whose zero-level set captures the underlying surface. To achieve the goal, existing methods rely on traditional meshing algorithms; while promising, they suffer from loss of precision learned in the implicit surface networks, due to the use of discrete space sampling in marching cubes. Given that an MLP with activations of Rectified Linear Unit (ReLU) partitions its input space into a number of linear regions, we are motivated to connect this local linearity with a same property owned by the desired result of polygon mesh. More specifically, we identify from the linear regions, partitioned by an MLP based implicit function, the analytic cells and analytic faces that are associated with the function's zero-level isosurface. We prove that under mild conditions, the identified analytic faces are guaranteed to connect and form a closed, piecewise planar surface. Based on the theorem, we propose an algorithm of analytic marching, which marches among analytic cells to exactly recover the mesh captured by an implicit surface network. We also show that our theory and algorithm are equally applicable to advanced MLPs with shortcut connections and max pooling. Given the parallel nature of analytic marching, we contribute AnalyticMesh, a software package that supports efficient meshing of implicit surface networks via CUDA parallel computing, and mesh simplification for efficient downstream processing. We apply our method to different settings of generative shape modeling using implicit surface networks. Extensive experiments demonstrate our advantages over existing methods in terms of both meshing accuracy and efficiency.

(2 min) In medical image segmentation, it is difficult to mark ambiguous areas accurately with binary masks, especially when dealing with small lesions. Therefore, it is a challenge for radiologists to reach a consensus by using binary masks under the condition of multiple annotations. However, these areas may contain anatomical structures that are conducive to diagnosis. Uncertainty is introduced to study these situations. Nevertheless, the uncertainty is usually measured by the variances between predictions in a multiple trial way. It is not intuitive, and there is no exact correspondence in the image. Inspired by image matting, we introduce matting as a soft segmentation method and a new perspective to deal with and represent uncertain regions into medical scenes, namely medical matting. More specifically, because there is no available medical matting dataset, we first labeled two medical datasets with alpha matte. Secondly, the matting method applied to the natural image is not suitable for the medical scene, so we propose a new architecture to generate binary masks and alpha matte in a row. Thirdly, the uncertainty map is introduced to highlight the ambiguous regions from the binary results and improve the matting performance. Evaluated on these datasets, the proposed model outperformed state-of-the-art matting algorithms by a large margin, and alpha matte is proved to be a more efficient labeling form than a binary mask.

(2 min) Fully autonomous drones have been demonstrated to find lost or injured persons under strongly occluding forest canopy. Airborne Optical Sectioning (AOS), a novel synthetic aperture imaging technique, together with deep-learning-based classification enables high detection rates under realistic search-and-rescue conditions. We demonstrate that false detections can be significantly suppressed and true detections boosted by combining classifications from multiple AOS rather than single integral images. This improves classification rates especially in the presence of occlusion. To make this possible, we modified the AOS imaging process to support large overlaps between subsequent integrals, enabling real-time and on-board scanning and processing of groundspeeds up to 10 m/s.

(2 min) This paper explores the potential for performing temporal semantic segmentation in the context of agricultural robotics without temporally labelled data. We achieve this by proposing to generate virtual temporal samples from labelled still images. This allows us, with no extra annotation effort, to generate virtually labelled temporal sequences. Normally, to train a recurrent neural network (RNN), labelled samples from a video (temporal) sequence are required which is laborious and has stymied work in this direction. By generating virtual temporal samples, we demonstrate that it is possible to train a lightweight RNN to perform semantic segmentation on two challenging agricultural datasets. Our results show that by training a temporal semantic segmenter using virtual samples we can increase the performance by an absolute amount of 4.6 and 4.9 on sweet pepper and sugar beet datasets, respectively. This indicates that our virtual data augmentation technique is able to accurately classify agricultural images temporally without the use of complicated synthetic data generation techniques nor with the overhead of labelling large amounts of temporal sequences.

(2 min) Fourier phase retrieval is the problem of reconstructing a signal given only the magnitude of its Fourier transformation. Optimization-based approaches, like the well-established Gerchberg-Saxton or the hybrid input output algorithm, struggle at reconstructing images from magnitudes that are not oversampled. This motivates the application of learned methods, which allow reconstruction from non-oversampled magnitude measurements after a learning phase. In this paper, we want to push the limits of these learned methods by means of a deep neural network cascade that reconstructs the image successively on different resolutions from its non-oversampled Fourier magnitude. We evaluate our method on four different datasets (MNIST, EMNIST, Fashion-MNIST, and KMNIST) and demonstrate that it yields improved performance over other non-iterative methods and optimization-based methods.

(2 min) Detecting generic, taxonomy-free event boundaries invideos represents a major stride forward towards holisticvideo understanding. In this paper we present a technique forgeneric event boundary detection based on a two stream in-flated 3D convolutions architecture, which can learn spatio-temporal features from videos. Our work is inspired from theGeneric Event Boundary Detection Challenge (part of CVPR2021 Long Form Video Understanding- LOVEU Workshop).Throughout the paper we provide an in-depth analysis ofthe experiments performed along with an interpretation ofthe results obtained.

(2 min) This work introduces World-GAN, the first method to perform data-driven Procedural Content Generation via Machine Learning in Minecraft from a single example. Based on a 3D Generative Adversarial Network (GAN) architecture, we are able to create arbitrarily sized world snippets from a given sample. We evaluate our approach on creations from the community as well as structures generated with the Minecraft World Generator. Our method is motivated by the dense representations used in Natural Language Processing (NLP) introduced with word2vec [1]. The proposed block2vec representations make World-GAN independent from the number of different blocks, which can vary a lot in Minecraft, and enable the generation of larger levels. Finally, we demonstrate that changing this new representation space allows us to change the generated style of an already trained generator. World-GAN enables its users to generate Minecraft worlds based on parts of their creations.

(2 min) In this report, we describe the technical details of our submission to the 2021 EPIC-KITCHENS-100 Unsupervised Domain Adaptation Challenge for Action Recognition. Leveraging multiple modalities has been proved to benefit the Unsupervised Domain Adaptation (UDA) task. In this work, we present Multi-Modal Mutual Enhancement Module (M3EM), a deep module for jointly considering information from multiple modalities to find the most transferable representations across domains. We achieve this by implementing two sub-modules for enhancing each modality using the context of other modalities. The first sub-module exchanges information across modalities through the semantic space, while the second sub-module finds the most transferable spatial region based on the consensus of all modalities.

(2 min) In this paper, we present GEM as a General Evaluation benchmark for Multimodal tasks. Different from existing datasets such as GLUE, SuperGLUE, XGLUE and XTREME that mainly focus on natural language tasks, GEM is a large-scale vision-language benchmark, which consists of GEM-I for image-language tasks and GEM-V for video-language tasks. Comparing with existing multimodal datasets such as MSCOCO and Flicker30K for image-language tasks, YouCook2 and MSR-VTT for video-language tasks, GEM is not only the largest vision-language dataset covering image-language tasks and video-language tasks at the same time, but also labeled in multiple languages. We also provide two baseline models for this benchmark. We will release the dataset, code and baseline models, aiming to advance the development of multilingual multimodal research.

(2 min) Pseudo-LiDAR based 3D object detectors have gained popularity due to their high accuracy. However, these methods need dense depth supervision and suffer from inferior speed. To solve these two issues, a recently introduced RTS3D builds an efficient 4D Feature-Consistency Embedding (FCE) space for the intermediate representation of object without depth supervision. FCE space splits the entire object region into 3D uniform grid latent space for feature sampling point generation, which ignores the importance of different object regions. However, we argue that, compared with the inner region, the outer region plays a more important role for accurate 3D detection. To encode more information from the outer region, we propose a shape prior non-uniform sampling strategy that performs dense sampling in outer region and sparse sampling in inner region. As a result, more points are sampled from the outer region and more useful features are extracted for 3D detection. Further, to enhance the feature discrimination of each sampling point, we propose a high-level semantic enhanced FCE module to exploit more contextual information and suppress noise better. Experiments on the KITTI dataset are performed to show the effectiveness of the proposed method. Compared with the baseline RTS3D, our proposed method has 2.57% improvement on AP3d almost without extra network parameters. Moreover, our proposed method outperforms the state-of-the-art methods without extra supervision at a real-time speed.

(2 min) In real-world image enhancement, it is often challenging (if not impossible) to acquire ground-truth data, preventing the adoption of distance metrics for objective quality assessment. As a result, one often resorts to subjective quality assessment, the most straightforward and reliable means of evaluating image enhancement. Conventional subjective testing requires manually pre-selecting a small set of visual examples, which may suffer from three sources of biases: 1) sampling bias due to the extremely sparse distribution of the selected samples in the image space; 2) algorithmic bias due to potential overfitting the selected samples; 3) subjective bias due to further potential cherry-picking test results. This eventually makes the field of real-world image enhancement more of an art than a science. Here we take steps towards debiasing conventional subjective assessment by automatically sampling a set of adaptive and diverse images for subsequent testing. This is achieved by casting sample selection into a joint maximization of the discrepancy between the enhancers and the diversity among the selected input images. Careful visual inspection on the resulting enhanced images provides a debiased ranking of the enhancement algorithms. We demonstrate our subjective assessment method using three popular and practically demanding image enhancement tasks: dehazing, super-resolution, and low-light enhancement.

(2 min) Interpreting how does deep neural networks (DNNs) make predictions is a vital field in artificial intelligence, which hinders wide applications of DNNs. Visualization of learned representations helps we humans understand the vision of DNNs. In this work, visualized images that can activate the neural network to the target classes are generated by back-propagation method. Here, rotation and scaling operations are applied to introduce the transformation invariance in the image generating process, which we find a significant improvement on visualization effect. Finally, we show some cases that such method can help us to gain insight into neural networks.

(2 min) Collecting training data from untrusted sources exposes machine learning services to poisoning adversaries, who maliciously manipulate training data to degrade the model accuracy. When trained on offline datasets, poisoning adversaries have to inject the poisoned data in advance before training, and the order of feeding these poisoned batches into the model is stochastic. In contrast, practical systems are more usually trained/fine-tuned on sequentially captured real-time data, in which case poisoning adversaries could dynamically poison each data batch according to the current model state. In this paper, we focus on the real-time settings and propose a new attacking strategy, which affiliates an accumulative phase with poisoning attacks to secretly (i.e., without affecting accuracy) magnify the destructive effect of a (poisoned) trigger batch. By mimicking online learning and federated learning on CIFAR-10, we show that the model accuracy will significantly drop by a single update step on the trigger batch after the accumulative phase. Our work validates that a well-designed but straightforward attacking strategy can dramatically amplify the poisoning effects, with no need to explore complex techniques.

(2 min) In this paper we present our preliminary work on model-based behavioral analysis of horse motion. Our approach is based on the SMAL model, a 3D articulated statistical model of animal shape. We define a novel SMAL model for horses based on a new template, skeleton and shape space learned from $37$ horse toys. We test the accuracy of our hSMAL model in reconstructing a horse from 3D mocap data and images. We apply the hSMAL model to the problem of lameness detection from video, where we fit the model to images to recover 3D pose and train an ST-GCN network on pose data. A comparison with the same network trained on mocap points illustrates the benefit of our approach.

(2 min) Training a Convolutional Neural Network (CNN) to be robust against rotation has mostly been done with data augmentation. In this paper, another progressive vision of research direction is highlighted to encourage less dependence on data augmentation by achieving structural rotational invariance of a network. The deep equivariance-bridged SO(2) invariant network is proposed to echo such vision. First, Self-Weighted Nearest Neighbors Graph Convolutional Network (SWN-GCN) is proposed to implement Graph Convolutional Network (GCN) on the graph representation of an image to acquire rotationally equivariant representation, as GCN is more suitable for constructing deeper network than spectral graph convolution-based approaches. Then, invariant representation is eventually obtained with Global Average Pooling (GAP), a permutation-invariant operation suitable for aggregating high-dimensional representations, over the equivariant set of vertices retrieved from SWN-GCN. Our method achieves the state-of-the-art image classification performance on rotated MNIST and CIFAR-10 images, where the models are trained with a non-augmented dataset only. Quantitative validations over invariance of the representations also demonstrate strong invariance of deep representations of SWN-GCN over rotations.

(2 min) In active visual tracking, it is notoriously difficult when distracting objects appear, as distractors often mislead the tracker by occluding the target or bringing a confusing appearance. To address this issue, we propose a mixed cooperative-competitive multi-agent game, where a target and multiple distractors form a collaborative team to play against a tracker and make it fail to follow. Through learning in our game, diverse distracting behaviors of the distractors naturally emerge, thereby exposing the tracker's weakness, which helps enhance the distraction-robustness of the tracker. For effective learning, we then present a bunch of practical methods, including a reward function for distractors, a cross-modal teacher-student learning strategy, and a recurrent attention mechanism for the tracker. The experimental results show that our tracker performs desired distraction-robust active visual tracking and can be well generalized to unseen environments. We also show that the multi-agent game can be used to adversarially test the robustness of trackers.

(2 min) Video affective understanding, which aims to predict the evoked expressions by the video content, is desired for video creation and recommendation. In the recent EEV challenge, a dense affective understanding task is proposed and requires frame-level affective prediction. In this paper, we propose a multi-granularity network with modal attention (MGN-MA), which employs multi-granularity features for better description of the target frame. Specifically, the multi-granularity features could be divided into frame-level, clips-level and video-level features, which corresponds to visual-salient content, semantic-context and video theme information. Then the modal attention fusion module is designed to fuse the multi-granularity features and emphasize more affection-relevant modals. Finally, the fused feature is fed into a Mixtures Of Experts (MOE) classifier to predict the expressions. Further employing model-ensemble post-processing, the proposed method achieves the correlation score of 0.02292 in the EEV challenge.

(2 min) Finding a suitable data representation for a specific task has been shown to be crucial in many applications. The success of subspace clustering depends on the assumption that the data can be separated into different subspaces. However, this simple assumption does not always hold since the raw data might not be separable into subspaces. To recover the ``clustering-friendly'' representation and facilitate the subsequent clustering, we propose a graph filtering approach by which a smooth representation is achieved. Specifically, it injects graph similarity into data features by applying a low-pass filter to extract useful data representations for clustering. Extensive experiments on image and document clustering datasets demonstrate that our method improves upon state-of-the-art subspace clustering techniques. Especially, its comparable performance with deep learning methods emphasizes the effectiveness of the simple graph filtering scheme for many real-world applications. An ablation study shows that graph filtering can remove noise, preserve structure in the image, and increase the separability of classes.

(2 min) Novelty detection is the process of determining whether a query example differs from the learned training distribution. Previous methods attempt to learn the representation of the normal samples via generative adversarial networks (GANs). However, they will suffer from instability training, mode dropping, and low discriminative ability. Recently, various pretext tasks (e.g. rotation prediction and clustering) have been proposed for self-supervised learning in novelty detection. However, the learned latent features are still low discriminative. We overcome such problems by introducing a novel decoder-encoder framework. Firstly, a generative network (a.k.a. decoder) learns the representation by mapping the initialized latent vector to an image. In particular, this vector is initialized by considering the entire distribution of training data to avoid the problem of mode-dropping. Secondly, a contrastive network (a.k.a. encoder) aims to ``learn to compare'' through mutual information estimation, which directly helps the generative network to obtain a more discriminative representation by using a negative data augmentation strategy. Extensive experiments show that our model has significant superiority over cutting-edge novelty detectors and achieves new state-of-the-art results on some novelty detection benchmarks, e.g. CIFAR10 and DCASE. Moreover, our model is more stable for training in a non-adversarial manner, compared to other adversarial based novelty detection methods.

(2 min) We present a heterogeneous localization framework for solving radar global localization and pose tracking on pre-built lidar maps. To bridge the gap of sensing modalities, deep neural networks are constructed to create shared embedding space for radar scans and lidar maps. Herein learned feature embeddings are supportive for similarity measurement, thus improving map retrieval and data matching respectively. In RobotCar and MulRan datasets, we demonstrate the effectiveness of the proposed framework with the comparison to Scan Context and RaLL. In addition, the proposed pose tracking pipeline is with less neural networks compared to the original RaLL.

(2 min) In this work, we propose a high fidelity face swapping method, called HifiFace, which can well preserve the face shape of the source face and generate photo-realistic results. Unlike other existing face swapping works that only use face recognition model to keep the identity similarity, we propose 3D shape-aware identity to control the face shape with the geometric supervision from 3DMM and 3D face reconstruction method. Meanwhile, we introduce the Semantic Facial Fusion module to optimize the combination of encoder and decoder features and make adaptive blending, which makes the results more photo-realistic. Extensive experiments on faces in the wild demonstrate that our method can preserve better identity, especially on the face shape, and can generate more photo-realistic results than previous state-of-the-art methods.

(2 min) Late gadolinium enhancement magnetic resonance imaging (LGE MRI) is commonly used to visualize and quantify left atrial (LA) scars. The position and extent of scars provide important information of the pathophysiology and progression of atrial fibrillation (AF). Hence, LA scar segmentation and quantification from LGE MRI can be useful in computer-assisted diagnosis and treatment stratification of AF patients. Since manual delineation can be time-consuming and subject to intra- and inter-expert variability, automating this computing is highly desired, which nevertheless is still challenging and under-researched. This paper aims to provide a systematic review on computing methods for LA cavity, wall, scar and ablation gap segmentation and quantification from LGE MRI, and the related literature for AF studies. Specifically, we first summarize AF-related imaging techniques, particularly LGE MRI. Then, we review the methodologies of the four computing tasks in detail, and summarize the validation strategies applied in each task. Finally, the possible future developments are outlined, with a brief survey on the potential clinical applications of the aforementioned methods. The review shows that the research into this topic is still in early stages. Although several methods have been proposed, especially for LA segmentation, there is still large scope for further algorithmic developments due to performance issues related to the high variability of enhancement appearance and differences in image acquisition.

(2 min) In a previous study, we presented VT-Lane, a three-step framework for real-time vehicle detection, tracking, and turn movement classification at urban intersections. In this study, we present a case study incorporating the highly accurate trajectories and movement classification obtained via VT-Lane for the purpose of speed estimation and driver behavior calibration for traffic at urban intersections. First, we use a highly instrumented vehicle to verify the estimated speeds obtained from video inference. The results of the speed validation show that our method can estimate the average travel speed of detected vehicles in real-time with an error of 0.19 m/sec, which is equivalent to 2% of the average observed travel speeds in the intersection of the study. Instantaneous speeds (at the resolution of 30 Hz) were found to be estimated with an average error of 0.21 m/sec and 0.86 m/sec respectively for free-flowing and congested traffic conditions. We then use the estimated speeds to calibrate the parameters of a driver behavior model for the vehicles in the area of study. The results show that the calibrated model replicates the driving behavior with an average error of 0.45 m/sec, indicating the high potential for using this framework for automated, large-scale calibration of car-following models from roadside traffic video data, which can lead to substantial improvements in traffic modeling via microscopic simulation.

(2 min) Limited availability of labeled-data makes any supervised learning problem challenging. Alternative learning settings like semi-supervised and universum learning alleviate the dependency on labeled data, but still require a large amount of unlabeled data, which may be unavailable or expensive to acquire. GAN-based synthetic data generation methods have recently shown promise by generating synthetic samples to improve task at hand. However, these samples cannot be used for other purposes. In this paper, we propose a GAN game which provides improved discriminator accuracy under limited data settings, while generating realistic synthetic data. This provides the added advantage that now the generated data can be used for other similar tasks. We provide the theoretical guarantees and empirical results in support of our approach.

(2 min) By the ALARA (As Low As Reasonably Achievable) principle, ultra-low-dose CT reconstruction is a holy grail to minimize cancer risks and genetic damages, especially for children. With the development of medical CT technologies, the iterative algorithms are widely used to reconstruct decent CT images from a low-dose scan. Recently, artificial intelligence (AI) techniques have shown a great promise in further reducing CT radiation dose to the next level. In this paper, we demonstrate that AI-powered CT reconstruction offers diagnostic image quality at an ultra-low-dose level comparable to that of radiography. Specifically, here we develop a Split Unrolled Grid-like Alternative Reconstruction (SUGAR) network, in which deep learning, physical modeling and image prior are integrated. The reconstruction results from clinical datasets show that excellent images can be reconstructed using SUGAR from 36 projections. This approach has a potential to change future healthcare.

(2 min) Deep neural networks (DNNs) are effective in solving many real-world problems. Larger DNN models usually exhibit better quality (e.g., accuracy) but their excessive computation results in long training and inference time. Model sparsification can reduce the computation and memory cost while maintaining model quality. Most existing sparsification algorithms unidirectionally remove weights, while others randomly or greedily explore a small subset of weights in each layer. The inefficiency of the algorithms reduces the achievable sparsity level. In addition, many algorithms still require pre-trained dense models and thus suffer from large memory footprint and long training time. In this paper, we propose a novel scheduled grow-and-prune (GaP) methodology without pre-training the dense models. It addresses the shortcomings of the previous works by repeatedly growing a subset of layers to dense and then pruning back to sparse after some training. Experiments have shown that such models can match or beat the quality of highly optimized dense models at 80% sparsity on a variety of tasks, such as image classification, objective detection, 3D object part segmentation, and translation. They also outperform other state-of-the-art (SOTA) pruning methods, including pruning from pre-trained dense models. As an example, a 90% sparse ResNet-50 obtained via GaP achieves 77.9% top-1 accuracy on ImageNet, improving the SOTA results by 1.5%.

(2 min) We tackle the problem of learning the geometry of multiple categories of deformable objects jointly. Recent work has shown that it is possible to learn a unified dense pose predictor for several categories of related objects. However, training such models requires to initialize inter-category correspondences by hand. This is suboptimal and the resulting models fail to maintain correct correspondences as individual categories are learned. In this paper, we show that improved correspondences can be learned automatically as a natural byproduct of learning category-specific dense pose predictors. To do this, we express correspondences between different categories and between images and categories using a unified embedding. Then, we use the latter to enforce two constraints: symmetric inter-category cycle consistency and a new asymmetric image-to-category cycle consistency. Without any manual annotations for the inter-category correspondences, we obtain state-of-the-art alignment results, outperforming dedicated methods for matching 3D shapes. Moreover, the new model is also better at the task of dense pose prediction than prior work.

(2 min) In this work we address the problem of landmark-based segmentation for anatomical structures. We propose HybridGNet, an encoder-decoder neural architecture which combines standard convolutions for image feature encoding, with graph convolutional neural networks to decode plausible representations of anatomical structures. We benchmark the proposed architecture considering other standard landmark and pixel-based models for anatomical segmentation in chest x-ray images, and found that HybridGNet is more robust to image occlusions. We also show that it can be used to construct landmark-based segmentations from pixel level annotations. Our experimental results suggest that HybridGNet produces accurate and anatomically plausible landmark-based segmentations, by naturally incorporating shape constraints within the decoding process via spectral convolutions.

(3 min) Purpose: Image classification is perhaps the most fundamental task in imaging AI. However, labeling images is time-consuming and tedious. We have recently demonstrated that reinforcement learning (RL) can classify 2D slices of MRI brain images with high accuracy. Here we make two important steps toward speeding image classification: Firstly, we automatically extract class labels from the clinical reports. Secondly, we extend our prior 2D classification work to fully 3D image volumes from our institution. Hence, we proceed as follows: in Part 1, we extract labels from reports automatically using the SBERT natural language processing approach. Then, in Part 2, we use these labels with RL to train a classification Deep-Q Network (DQN) for 3D image volumes. Methods: For Part 1, we trained SBERT with 90 radiology report impressions. We then used the trained SBERT to predict class labels for use in Part 2. In Part 2, we applied multi-step image classification to allow for combined Deep-Q learning using 3D convolutions and TD(0) Q learning. We trained on a set of 90 images. We tested on a separate set of 61 images, again using the classes predicted from patient reports by the trained SBERT in Part 1. For comparison, we also trained and tested a supervised deep learning classification network on the same set of training and testing images using the same labels. Results: Part 1: Upon training with the corpus of radiology reports, the SBERT model had 100% accuracy for both normal and metastasis-containing scans. Part 2: Then, using these labels, whereas the supervised approach quickly overfit the training data and as expected performed poorly on the testing set (66% accuracy, just over random guessing), the reinforcement learning approach achieved an accuracy of 92%. The results were found to be statistically significant, with a p-value of 3.1 x 10^-5.

(2 min) Imbalanced datasets widely exist in practice and area great challenge for training deep neural models with agood generalization on infrequent classes. In this work, wepropose a new rare-class sample generator (RSG) to solvethis problem. RSG aims to generate some new samplesfor rare classes during training, and it has in particularthe following advantages: (1) it is convenient to use andhighly versatile, because it can be easily integrated intoany kind of convolutional neural network, and it works wellwhen combined with different loss functions, and (2) it isonly used during the training phase, and therefore, no ad-ditional burden is imposed on deep neural networks duringthe testing phase. In extensive experimental evaluations, weverify the effectiveness of RSG. Furthermore, by leveragingRSG, we obtain competitive results on Imbalanced CIFARand new state-of-the-art results on Places-LT, ImageNet-LT, and iNaturalist 2018. The source code is available at https://github.com/Jianf-Wang/RSG.

(2 min) Evaluating robustness of machine-learning models to adversarial examples is a challenging problem. Many defenses have been shown to provide a false sense of security by causing gradient-based attacks to fail, and they have been broken under more rigorous evaluations. Although guidelines and best practices have been suggested to improve current adversarial robustness evaluations, the lack of automatic testing and debugging tools makes it difficult to apply these recommendations in a systematic manner. In this work, we overcome these limitations by (i) defining a set of quantitative indicators which unveil common failures in the optimization of gradient-based attacks, and (ii) proposing specific mitigation strategies within a systematic evaluation protocol. Our extensive experimental analysis shows that the proposed indicators of failure can be used to visualize, debug and improve current adversarial robustness evaluations, providing a first concrete step towards automatizing and systematizing current adversarial robustness evaluations. Our open-source code is available at: https://github.com/pralab/IndicatorsOfAttackFailure.

(2 min) We propose a novel GAN training scheme that can handle any level of labeling in a unified manner. Our scheme introduces a form of artificial labeling that can incorporate manually defined labels, when available, and induce an alignment between them. To define the artificial labels, we exploit the assumption that neural network generators can be trained more easily to map nearby latent vectors to data with semantic similarities, than across separate categories. We use generated data samples and their corresponding artificial conditioning labels to train a classifier. The classifier is then used to self-label real data. To boost the accuracy of the self-labeling, we also use the exponential moving average of the classifier. However, because the classifier might still make mistakes, especially at the beginning of the training, we also refine the labels through self-attention, by using the labeling of real data samples only when the classifier outputs a high classification probability score. We evaluate our approach on CIFAR-10, STL-10 and SVHN, and show that both self-labeling and self-attention consistently improve the quality of generated data. More surprisingly, we find that the proposed scheme can even outperform class-conditional GANs.

(2 min) Integrated Gradients (IG) is a commonly used feature attribution method for deep neural networks. While IG has many desirable properties, the method often produces spurious/noisy pixel attributions in regions that are not related to the predicted class when applied to visual models. While this has been previously noted, most existing solutions are aimed at addressing the symptoms by explicitly reducing the noise in the resulting attributions. In this work, we show that one of the causes of the problem is the accumulation of noise along the IG path. To minimize the effect of this source of noise, we propose adapting the attribution path itself -- conditioning the path not just on the image but also on the model being explained. We introduce Adaptive Path Methods (APMs) as a generalization of path methods, and Guided IG as a specific instance of an APM. Empirically, Guided IG creates saliency maps better aligned with the model's prediction and the input image that is being explained. We show through qualitative and quantitative experiments that Guided IG outperforms other, related methods in nearly every experiment.

(2 min) DeepLab2 is a TensorFlow library for deep labeling, aiming to provide a state-of-the-art and easy-to-use TensorFlow codebase for general dense pixel prediction problems in computer vision. DeepLab2 includes all our recently developed DeepLab model variants with pretrained checkpoints as well as model training and evaluation code, allowing the community to reproduce and further improve upon the state-of-art systems. To showcase the effectiveness of DeepLab2, our Panoptic-DeepLab employing Axial-SWideRNet as network backbone achieves 68.0% PQ or 83.5% mIoU on Cityscaspes validation set, with only single-scale inference and ImageNet-1K pretrained checkpoints. We hope that publicly sharing our library could facilitate future research on dense pixel labeling tasks and envision new applications of this technology. Code is made publicly available at \url{https://github.com/google-research/deeplab2}.

(2 min) The vulnerability of deep neural networks to adversarial examples, which are crafted maliciously by modifying the inputs with imperceptible perturbations to misled the network produce incorrect outputs, reveals the lack of robustness and poses security concerns. Previous works study the adversarial robustness of image classifiers on image level and use all the pixel information in an image indiscriminately, lacking of exploration of regions with different semantic meanings in the pixel space of an image. In this work, we fill this gap and explore the pixel space of the adversarial image by proposing an algorithm to looking for possible perturbations pixel by pixel in different regions of the segmented image. The extensive experimental results on CIFAR-10 and ImageNet verify that searching for the modified pixel in only some pixels of an image can successfully launch the one-pixel adversarial attacks without requiring all the pixels of the entire image, and there exist multiple vulnerable points scattered in different regions of an image. We also demonstrate that the adversarial robustness of different regions on the image varies with the amount of semantic information contained.

(2 min) The training of deep neural networks (DNNs) always requires intensive resources for both computation and data storage. Thus, DNNs cannot be efficiently applied to mobile phones and embedded devices, which severely limits their applicability in industrial applications. To address this issue, we propose a novel encoding scheme using {-1, +1} to decompose quantized neural networks (QNNs) into multi-branch binary networks, which can be efficiently implemented by bitwise operations (i.e., xnor and bitcount) to achieve model compression, computational acceleration, and resource saving. By using our method, users can achieve different encoding precisions arbitrarily according to their requirements and hardware resources. The proposed mechanism is highly suitable for the use of FPGA and ASIC in terms of data storage and computation, which provides a feasible idea for smart chips. We validate the effectiveness of our method on large-scale image classification (e.g., ImageNet), object detection, and semantic segmentation tasks. In particular, our method with low-bit encoding can still achieve almost the same performance as its high-bit counterparts.

(2 min) To evaluate clustering results is a significant part of cluster analysis. Since there are no true class labels for clustering in typical unsupervised learning, many internal cluster validity indices (CVIs), which use predicted labels and data, have been created. Without true labels, to design an effective CVI is as difficult as to create a clustering method. And it is crucial to have more CVIs because there are no universal CVIs that can be used to measure all datasets and no specific methods of selecting a proper CVI for clusters without true labels. Therefore, to apply a variety of CVIs to evaluate clustering results is necessary. In this paper, we propose a novel internal CVI -- the Distance-based Separability Index (DSI), based on a data separability measure. We compared the DSI with eight internal CVIs including studies from early Dunn (1974) to most recent CVDD (2019) and an external CVI as ground truth, by using clustering results of five clustering algorithms on 12 real and 97 synthetic datasets. Results show DSI is an effective, unique, and competitive CVI to other compared CVIs. We also summarized the general process to evaluate CVIs and created the rank-difference metric for comparison of CVIs' results.

(2 min) Machine learning is a general-purpose technology holding promises for many interdisciplinary research problems. However, significant barriers exist in crossing disciplinary boundaries when most machine learning tools are developed in different areas separately. We present Pykale - a Python library for knowledge-aware machine learning on graphs, images, texts, and videos to enable and accelerate interdisciplinary research. We formulate new green machine learning guidelines based on standard software engineering practices and propose a novel pipeline-based application programming interface (API). PyKale focuses on leveraging knowledge from multiple sources for accurate and interpretable prediction, thus supporting multimodal learning and transfer learning (particularly domain adaptation) with latest deep learning and dimensionality reduction models. We build PyKale on PyTorch and leverage the rich PyTorch ecosystem. Our pipeline-based API design enforces standardization and minimalism, embracing green machine learning concepts via reducing repetitions and redundancy, reusing existing resources, and recycling learning models across areas. We demonstrate its interdisciplinary nature via examples in bioinformatics, knowledge graph, image/video recognition, and medical imaging.

(2 min) This paper investigates two techniques for developing efficient self-supervised vision transformers (EsViT) for visual representation learning. First, we show through a comprehensive empirical study that multi-stage architectures with sparse self-attentions can significantly reduce modeling complexity but with a cost of losing the ability to capture fine-grained correspondences between image regions. Second, we propose a new pre-training task of region matching which allows the model to capture fine-grained region dependencies and as a result significantly improves the quality of the learned vision representations. Our results show that combining the two techniques, EsViT achieves 81.3% top-1 on the ImageNet linear probe evaluation, outperforming prior arts with around an order magnitude of higher throughput. When transferring to downstream linear classification tasks, EsViT outperforms its supervised counterpart on 17 out of 18 datasets. The code and models will be publicly available.

(2 min) A significant amount of work has been done on adversarial attacks that inject imperceptible noise to images to deteriorate the image classification performance of deep models. However, most of the existing studies consider attacks in the digital (pixel) domain where an image acquired by an image sensor with sampling and quantization has been recorded. This paper, for the first time, introduces an optical adversarial attack, which physically alters the light field information arriving at the image sensor so that the classification model yields misclassification. More specifically, we modulate the phase of the light in the Fourier domain using a spatial light modulator placed in the photographic system. The operative parameters of the modulator are obtained by gradient-based optimization to maximize cross-entropy and minimize distortions. We present experiments based on both simulation and a real hardware optical system, from which the feasibility of the proposed optical attack is demonstrated. It is also verified that the proposed attack is completely different from common optical-domain distortions such as spherical aberration, defocus, and astigmatism in terms of both perturbation patterns and classification results.

(2 min) We introduce a new dataset called Synthetic COVID-19 Chest X-ray Dataset for training machine learning models. The dataset consists of 21,295 synthetic COVID-19 chest X-ray images to be used for computer-aided diagnosis. These images, generated via an unsupervised domain adaptation approach, are of high quality. We find that the synthetic images not only improve performance of various deep learning architectures when used as additional training data under heavy imbalance conditions, but also detect the target class with high confidence. We also find that comparable performance can also be achieved when trained only on synthetic images. Further, salient features of the synthetic COVID-19 images indicate that the distribution is significantly different from Non-COVID-19 classes, enabling a proper decision boundary. We hope the availability of such high fidelity chest X-ray images of COVID-19 will encourage advances in the development of diagnostic and/or management tools.

(2 min) In recent years, multi-view subspace clustering has achieved impressive performance due to the exploitation of complementary imformation across multiple views. However, multi-view data can be very complicated and are not easy to cluster in real-world applications. Most existing methods operate on raw data and may not obtain the optimal solution. In this work, we propose a novel multi-view clustering method named smoothed multi-view subspace clustering (SMVSC) by employing a novel technique, i.e., graph filtering, to obtain a smooth representation for each view, in which similar data points have similar feature values. Specifically, it retains the graph geometric features through applying a low-pass filter. Consequently, it produces a ``clustering-friendly" representation and greatly facilitates the downstream clustering task. Extensive experiments on benchmark datasets validate the superiority of our approach. Analysis shows that graph filtering increases the separability of classes.

(2 min) This paper proposes two novel knowledge transfer techniques for class-incremental learning (CIL). First, we propose data-free generative replay (DF-GR) to mitigate catastrophic forgetting in CIL by using synthetic samples from a generative model. In the conventional generative replay, the generative model is pre-trained for old data and shared in extra memory for later incremental learning. In our proposed DF-GR, we train a generative model from scratch without using any training data, based on the pre-trained classification model from the past, so we curtail the cost of sharing pre-trained generative models. Second, we introduce dual-teacher information distillation (DT-ID) for knowledge distillation from two teachers to one student. In CIL, we use DT-ID to learn new classes incrementally based on the pre-trained model for old classes and another model (pre-)trained on the new data for new classes. We implemented the proposed schemes on top of one of the state-of-the-art CIL methods and showed the performance improvement on CIFAR-100 and ImageNet datasets.

(2 min) Query by Example is a well-known information retrieval task in which a document is chosen by the user as the search query and the goal is to retrieve relevant documents from a large collection. However, a document often covers multiple aspects of a topic. To address this scenario we introduce the task of faceted Query by Example in which users can also specify a finer grained aspect in addition to the input query document. We focus on the application of this task in scientific literature search. We envision models which are able to retrieve scientific papers analogous to a query scientific paper along specifically chosen rhetorical structure elements as one solution to this problem. In this work, the rhetorical structure elements, which we refer to as facets, indicate backgrounds, methods, or results of a scientific paper. We introduce and describe an expert annotated test collection to evaluate models trained to perform this task. Our test collection consists of a diverse set of 50 query documents, drawn from computational linguistics and machine learning venues. We carefully followed the annotation guideline used by TREC for depth-k pooling (k = 100 or 250) and the resulting data collection consists of graded relevance scores with high annotation agreement. The data is freely available for research purposes.

(2 min) Representation learning on user-item graph for recommendation has evolved from using single ID or interaction history to exploiting higher-order neighbors. This leads to the success of graph convolution networks (GCNs) for recommendation such as PinSage and LightGCN. Despite effectiveness, we argue that they suffer from two limitations: (1) high-degree nodes exert larger impact on the representation learning, deteriorating the recommendations of low-degree (long-tail) items; and (2) representations are vulnerable to noisy interactions, as the neighborhood aggregation scheme further enlarges the impact of observed edges. In this work, we explore self-supervised learning on user-item graph, so as to improve the accuracy and robustness of GCNs for recommendation. The idea is to supplement the classical supervised task of recommendation with an auxiliary self-supervised task, which reinforces node representation learning via self-discrimination. Specifically, we generate multiple views of a node, maximizing the agreement between different views of the same node compared to that of other nodes. We devise three operators to generate the views -- node dropout, edge dropout, and random walk -- that change the graph structure in different manners. We term this new learning paradigm as \textit{Self-supervised Graph Learning} (SGL), implementing it on the state-of-the-art model LightGCN. Through theoretical analyses, we find that SGL has the ability of automatically mining hard negatives. Empirical studies on three benchmark datasets demonstrate the effectiveness of SGL, which improves the recommendation accuracy, especially on long-tail items, and the robustness against interaction noises. Our implementations are available at \url{https://github.com/wujcan/SGL}.

(2 min) Pre-training and fine-tuning have achieved remarkable success in many downstream natural language processing (NLP) tasks. Recently, pre-training methods tailored for information retrieval (IR) have also been explored, and the latest success is the PROP method which has reached new SOTA on a variety of ad-hoc retrieval benchmarks. The basic idea of PROP is to construct the \textit{representative words prediction} (ROP) task for pre-training inspired by the query likelihood model. Despite its exciting performance, the effectiveness of PROP might be bounded by the classical unigram language model adopted in the ROP task construction process. To tackle this problem, we propose a bootstrapped pre-training method (namely B-PROP) based on BERT for ad-hoc retrieval. The key idea is to use the powerful contextual language model BERT to replace the classical unigram language model for the ROP task construction, and re-train BERT itself towards the tailored objective for IR. Specifically, we introduce a novel contrastive method, inspired by the divergence-from-randomness idea, to leverage BERT's self-attention mechanism to sample representative words from the document. By further fine-tuning on downstream ad-hoc retrieval tasks, our method achieves significant improvements over baselines without pre-training or with other pre-training methods, and further pushes forward the SOTA on a variety of ad-hoc retrieval tasks.

(2 min) Building a benchmark dataset for hate speech detection presents several challenges. Firstly, because hate speech is relatively rare -- e.g., less than 3\% of Twitter posts are hateful \citep{founta2018large} -- random sampling of tweets to annotate is inefficient in capturing hate speech. A common practice is to only annotate tweets containing known ``hate words'', but this risks yielding a biased benchmark that only partially captures the real-world phenomenon of interest. A second challenge is that definitions of hate speech tend to be highly variable and subjective. Annotators having diverse prior notions of hate speech may not only disagree with one another but also struggle to conform to specified labeling guidelines. Our key insight is that the rarity and subjectivity of hate speech are akin to that of relevance in information retrieval (IR). This connection suggests that well-established methodologies for creating IR test collections might also be usefully applied to create better benchmark datasets for hate speech detection. Firstly, to intelligently and efficiently select which tweets to annotate, we apply established IR techniques of {\em pooling} and {\em active learning}. Secondly, to improve both consistency and value of annotations, we apply {\em task decomposition} \cite{Zhang-sigir14} and {\em annotator rationale} \cite{mcdonnell16-hcomp} techniques. Using the above techniques, we create and share a new benchmark dataset\footnote{We will release the dataset upon publication.} for hate speech detection with broader coverage than prior datasets. We also show a dramatic drop in accuracy of existing detection models when tested on these broader forms of hate. Collected annotator rationales not only provide documented support for labeling decisions but also create exciting future work opportunities for dual-supervision and/or explanation generation in modeling.

(2 min) Point-of-Interest recommendation is an increasing research and developing area within the widely adopted technologies known as Recommender Systems. Among them, those that exploit information coming from Location-Based Social Networks (LBSNs) are very popular nowadays and could work with different information sources, which pose several challenges and research questions to the community as a whole. We present a systematic review focused on the research done in the last 10 years about this topic. We discuss and categorize the algorithms and evaluation methodologies used in these works and point out the opportunities and challenges that remain open in the field. More specifically, we report the leading recommendation techniques and information sources that have been exploited more often (such as the geographical signal and deep learning approaches) while we also alert about the lack of reproducibility in the field that may hinder real performance improvements.

(2 min) Technology-assisted review (TAR) refers to human-in-the-loop active learning workflows for finding relevant documents in large collections. These workflows often must meet a target for the proportion of relevant documents found (i.e. recall) while also holding down costs. A variety of heuristic stopping rules have been suggested for striking this tradeoff in particular settings, but none have been tested against a range of recall targets and tasks. We propose two new heuristic stopping rules, Quant and QuantCI based on model-based estimation techniques from survey research. We compare them against a range of proposed heuristics and find they are accurate at hitting a range of recall targets while substantially reducing review costs.

(2 min) Financial technology (FinTech) has drawn much attention among investors and companies. While conventional stock analysis in FinTech targets at predicting stock prices, less effort is made for profitable stock recommendation. Besides, in existing approaches on modeling time series of stock prices, the relationships among stocks and sectors (i.e., categories of stocks) are either neglected or pre-defined. Ignoring stock relationships will miss the information shared between stocks while using pre-defined relationships cannot depict the latent interactions or influence of stock prices between stocks. In this work, we aim at recommending the top-K profitable stocks in terms of return ratio using time series of stock prices and sector information. We propose a novel deep learning-based model, Financial Graph Attention Networks (FinGAT), to tackle the task under the setting that no pre-defined relationships between stocks are given. The idea of FinGAT is three-fold. First, we devise a hierarchical learning component to learn short-term and long-term sequential patterns from stock time series. Second, a fully-connected graph between stocks and a fully-connected graph between sectors are constructed, along with graph attention networks, to learn the latent interactions among stocks and sectors. Third, a multi-task objective is devised to jointly recommend the profitable stocks and predict the stock movement. Experiments conducted on Taiwan Stock, S&P 500, and NASDAQ datasets exhibit remarkable recommendation performance of our FinGAT, comparing to state-of-the-art methods.

(2 min) Technology-assisted review (TAR) refers to human-in-the-loop machine learning workflows for document review in legal discovery and other high recall review tasks. Attorneys and legal technologists have debated whether review should be a single iterative process (one-phase TAR workflows) or whether model training and review should be separate (two-phase TAR workflows), with implications for the choice of active learning algorithm. The relative cost of manual labeling for different purposes (training vs. review) and of different documents (positive vs. negative examples) is a key and neglected factor in this debate. Using a novel cost dynamics analysis, we show analytically and empirically that these relative costs strongly impact whether a one-phase or two-phase workflow minimizes cost. We also show how category prevalence, classification task difficulty, and collection size impact the optimal choice not only of workflow type, but of active learning method and stopping point.

(2 min) We propose the novel use of a generative adversarial network (GAN) (i) to make predictions in time (PredGAN) and (ii) to assimilate measurements (DA-PredGAN). In the latter case, we take advantage of the natural adjoint-like properties of generative models and the ability to simulate forwards and backwards in time. GANs have received much attention recently, after achieving excellent results for their generation of realistic-looking images. We wish to explore how this property translates to new applications in computational modelling and to exploit the adjoint-like properties for efficient data assimilation. To predict the spread of COVID-19 in an idealised town, we apply these methods to a compartmental model in epidemiology that is able to model space and time variations. To do this, the GAN is set within a reduced-order model (ROM), which uses a low-dimensional space for the spatial distribution of the simulation states. Then the GAN learns the evolution of the low-dimensional states over time. The results show that the proposed methods can accurately predict the evolution of the high-fidelity numerical simulation, and can efficiently assimilate observed data and determine the corresponding model parameters.

(2 min) Modeling the time-series of high-dimensional, longitudinal data is important for predicting patient disease progression. However, existing neural network based approaches that learn representations of patient state, while very flexible, are susceptible to overfitting. We propose a deep generative model that makes use of a novel attention-based neural architecture inspired by the physics of how treatments affect disease state. The result is a scalable and accurate model of high-dimensional patient biomarkers as they vary over time. Our proposed model yields significant improvements in generalization and, on real-world clinical data, provides interpretable insights into the dynamics of cancer progression.

(2 min) Despite its best performance in image denoising, the supervised deep denoising methods require paired noise-clean data, which are often unavailable. To address this challenge, Noise2Noise was designed based on the fact that paired noise-clean images can be replaced by paired noise-noise images that are easier to collect. However, in many scenarios the collection of paired noise-noise images is still impractical. To bypass labeled images, Noise2Void methods predict masked pixels from their surroundings with single noisy images only and give improved denoising results that still need improvements. An observation on classic denoising methods is that non-local mean (NLM) outcomes are typically superior to locally denoised results. In contrast, Noise2Void and its variants do not utilize self-similarities in an image as the NLM-based methods do. Here we propose Noise2Sim, an NLM-inspired self-learning method for image denoising. Specifically, Noise2Sim leverages the self-similarity of image pixels to train the denoising network, requiring single noisy images only. Our theoretical analysis shows that Noise2Sim tends to be equivalent to Noise2Noise under mild conditions. To efficiently manage the computational burden for globally searching similar pixels, we design a two-step procedure to provide data for Noise2Sim training. Extensive experiments demonstrate the superiority of Noise2Sim on common benchmark datasets.

(2 min) We consider a decision maker who must choose an action in order to maximize a reward function that depends also on an unknown parameter {\Theta}. The decision maker can delay taking the action in order to experiment and gather additional information on {\Theta}. We model the decision maker's problem using a Bayesian sequential experimentation framework and use dynamic programming and diffusion-asymptotic analysis to solve it. For that, we scale our problem in a way that both the average number of experiments that is conducted per unit of time is large and the informativeness of each individual experiment is low. Under such regime, we derive a diffusion approximation for the sequential experimentation problem, which provides a number of important insights about the nature of the problem and its solution. Our solution method also shows that the complexity of the problem grows only quadratically with the cardinality of the set of actions from which the decision maker can choose. We illustrate our methodology and results using a concrete application in the context of assortment selection and new product introduction. Specifically, we study the problem of a seller who wants to select an optimal assortment of products to launch into the marketplace and is uncertain about consumers' preferences. Motivated by emerging practices in e-commerce, we assume that the seller is able to use a crowdvoting system to learn these preferences before a final assortment decision is made. In this context, we undertake an extensive numerical analysis to assess the value of learning and demonstrate the effectiveness and robustness of the heuristics derived from the diffusion approximation.

(2 min) Most environmental data come from a minority of well-monitored sites. An ongoing challenge in the environmental sciences is transferring knowledge from monitored sites to unmonitored sites. Here, we demonstrate a novel transfer learning framework that accurately predicts depth-specific temperature in unmonitored lakes (targets) by borrowing models from well-monitored lakes (sources). This method, Meta Transfer Learning (MTL), builds a meta-learning model to predict transfer performance from candidate source models to targets using lake attributes and candidates' past performance. We constructed source models at 145 well-monitored lakes using calibrated process-based modeling (PB) and a recently developed approach called process-guided deep learning (PGDL). We applied MTL to either PB or PGDL source models (PB-MTL or PGDL-MTL, respectively) to predict temperatures in 305 target lakes treated as unmonitored in the Upper Midwestern United States. We show significantly improved performance relative to the uncalibrated process-based General Lake Model, where the median RMSE for the target lakes is $2.52^{\circ}C$. PB-MTL yielded a median RMSE of $2.43^{\circ}C$; PGDL-MTL yielded $2.16^{\circ}C$; and a PGDL-MTL ensemble of nine sources per target yielded $1.88^{\circ}C$. For sparsely monitored target lakes, PGDL-MTL often outperformed PGDL models trained on the target lakes themselves. Differences in maximum depth between the source and target were consistently the most important predictors. Our approach readily scales to thousands of lakes in the Midwestern United States, demonstrating that MTL with meaningful predictor variables and high-quality source models is a promising approach for many kinds of unmonitored systems and environmental variables.

(2 min) Approximate inference in deep Bayesian networks exhibits a dilemma of how to yield high fidelity posterior approximations while maintaining computational efficiency and scalability. We tackle this challenge by introducing a novel variational structured approximation inspired by the Bayesian interpretation of Dropout regularization. Concretely, we focus on the inflexibility of the factorized structure in Dropout posterior and then propose an improved method called Variational Structured Dropout (VSD). VSD employs an orthogonal transformation to learn a structured representation on the variational noise and consequently induces statistical dependencies in the approximate posterior. Theoretically, VSD successfully addresses the pathologies of previous Variational Dropout methods and thus offers a standard Bayesian justification. We further show that VSD induces an adaptive regularization term with several desirable properties which contribute to better generalization. Finally, we conduct extensive experiments on standard benchmarks to demonstrate the effectiveness of VSD over state-of-the-art variational methods on predictive accuracy, uncertainty estimation, and out-of-distribution detection.

(2 min) In adversarial training (AT), the main focus has been the objective and optimizer while the model has been less studied, so that the models being used are still those classic ones in standard training (ST). Classic network architectures (NAs) are generally worse than searched NAs in ST, which should be the same in AT. In this paper, we argue that NA and AT cannot be handled independently, since given a dataset, the optimal NA in ST would be no longer optimal in AT. That being said, AT is time-consuming itself; if we directly search NAs in AT over large search spaces, the computation will be practically infeasible. Thus, we propose a diverse-structured network (DS-Net), to significantly reduce the size of the search space: instead of low-level operations, we only consider predefined atomic blocks, where an atomic block is a time-tested building block like the residual block. There are only a few atomic blocks and thus we can weight all atomic blocks rather than find the best one in a searched block of DS-Net, which is an essential trade-off between exploring diverse structures and exploiting the best structures. Empirical results demonstrate the advantages of DS-Net, i.e., weighting the atomic blocks.

(2 min) Decentralized training of deep learning models enables on-device learning over networks, as well as efficient scaling to large compute clusters. Experiments in earlier works reveal that, even in a data-center setup, decentralized training often suffers from the degradation in the quality of the model: the training and test performance of models trained in a decentralized fashion is in general worse than that of models trained in a centralized fashion, and this performance drop is impacted by parameters such as network size, communication topology and data partitioning. We identify the changing consensus distance between devices as a key parameter to explain the gap between centralized and decentralized training. We show in theory that when the training consensus distance is lower than a critical quantity, decentralized training converges as fast as the centralized counterpart. We empirically validate that the relation between generalization performance and consensus distance is consistent with this theoretical observation. Our empirical insights allow the principled design of better decentralized training schemes that mitigate the performance drop. To this end, we provide practical training guidelines and exemplify its effectiveness on the data-center setup as the important first step.

(2 min) Machine Learning models have been deployed across many different aspects of society, often in situations that affect social welfare. Although these models offer streamlined solutions to large problems, they may contain biases and treat groups or individuals unfairly based on protected attributes such as gender. In this paper, we introduce several examples of machine learning gender bias in practice followed by formalizations of fairness. We provide a survey of fairness research by detailing influential pre-processing, in-processing, and post-processing bias mitigation algorithms. We then propose an \textup{end-to-end bias mitigation} framework, which employs a fusion of pre-, in-, and post-processing methods to leverage the strengths of each individual technique. We test this method, along with the standard techniques we review, on a deep neural network to analyze bias mitigation in a deep learning setting. We find that our end-to-end bias mitigation framework outperforms the baselines with respect to several fairness metrics, suggesting its promise as a method for improving fairness. As society increasingly relies on artificial intelligence to help in decision-making, addressing gender biases present in deep learning models is imperative. To provide readers with the tools to assess the fairness of machine learning models and mitigate the biases present in them, we discuss multiple open source packages for fairness in AI.

(2 min) We present a plug-in replacement for batch normalization (BN) called exponential moving average normalization (EMAN), which improves the performance of existing student-teacher based self- and semi-supervised learning techniques. Unlike the standard BN, where the statistics are computed within each batch, EMAN, used in the teacher, updates its statistics by exponential moving average from the BN statistics of the student. This design reduces the intrinsic cross-sample dependency of BN and enhances the generalization of the teacher. EMAN improves strong baselines for self-supervised learning by 4-6/1-2 points and semi-supervised learning by about 7/2 points, when 1%/10% supervised labels are available on ImageNet. These improvements are consistent across methods, network architectures, training duration, and datasets, demonstrating the general effectiveness of this technique. The code is available at https://github.com/amazon-research/exponential-moving-average-normalization.

(2 min) Implicit neural networks, a.k.a., deep equilibrium networks, are a class of implicit-depth learning models where function evaluation is performed by solving a fixed point equation. They generalize classic feedforward models and are equivalent to infinite-depth weight-tied feedforward networks. While implicit models show improved accuracy and significant reduction in memory consumption, they can suffer from ill-posedness and convergence instability. This paper provides a new framework to design well-posed and robust implicit neural networks based upon contraction theory for the non-Euclidean norm $\ell_\infty$. Our framework includes (i) a novel condition for well-posedness based on one-sided Lipschitz constants, (ii) an average iteration for computing fixed-points, and (iii) explicit estimates on input-output Lipschitz constants. Additionally, we design a training problem with the well-posedness condition and the average iteration as constraints and, to achieve robust models, with the input-output Lipschitz constant as a regularizer. Our $\ell_\infty$ well-posedness condition leads to a larger polytopic training search space than existing conditions and our average iteration enjoys accelerated convergence. Finally, we perform several numerical experiments for function estimation and digit classification through the MNIST data set. Our numerical results demonstrate improved accuracy and robustness of the implicit models with smaller input-output Lipschitz bounds.

(2 min) Machine learning technology has the potential to dramatically optimise event generation and simulations. We continue to investigate the use of neural networks to approximate matrix elements for high-multiplicity scattering processes. We focus on the case of loop-induced diphoton production through gluon fusion and develop a realistic simulation method that can be applied to hadron collider observables. Neural networks are trained using the one-loop amplitudes implemented in the NJet C++ library and interfaced to the Sherpa Monte Carlo event generator where we perform a detailed study for $2\to3$ and $2\to4$ scattering problems. We also consider how the trained networks perform when varying the kinematic cuts effecting the phase space and the reliability of the neural network simulations.

(2 min) Bootstrapping speech recognition on limited data resources has been an area of active research for long. The recent transition to all-neural models and end-to-end (E2E) training brought along particular challenges as these models are known to be data hungry, but also came with opportunities around language-agnostic representations derived from multilingual data as well as shared word-piece output representations across languages that share script and roots. We investigate here the effectiveness of different strategies to bootstrap an RNN-Transducer (RNN-T) based automatic speech recognition (ASR) system in the low resource regime, while exploiting the abundant resources available in other languages as well as the synthetic audio from a text-to-speech (TTS) engine. Our experiments demonstrate that transfer learning from a multilingual model, using a post-ASR text-to-text mapping and synthetic audio deliver additive improvements, allowing us to bootstrap a model for a new language with a fraction of the data that would otherwise be needed. The best system achieved a 46% relative word error rate (WER) reduction compared to the monolingual baseline, among which 25% relative WER improvement is attributed to the post-ASR text-to-text mappings and the TTS synthetic data.

(2 min) We propose a new approach to increase inference performance in environments that require a specific sequence of actions in order to be solved. This is for example the case for maze environments where ideally an optimal path is determined. Instead of learning a policy for a single step, we want to learn a policy that can predict n actions in advance. Our proposed method called policy horizon regression (PHR) uses knowledge of the environment sampled by A2C to learn an n dimensional policy vector in a policy distillation setup which yields n sequential actions per observation. We test our method on the MiniGrid and Pong environments and show drastic speedup during inference time by successfully predicting sequences of actions on a single observation.

(2 min) Tree-based models are among the most efficient machine learning techniques for data mining nowadays due to their accuracy, interpretability, and simplicity. The recent orthogonal needs for more data and privacy protection call for collaborative privacy-preserving solutions. In this work, we survey the literature on distributed and privacy-preserving training of tree-based models and we systematize its knowledge based on four axes: the learning algorithm, the collaborative model, the protection mechanism, and the threat model. We use this to identify the strengths and limitations of these works and provide for the first time a framework analyzing the information leakage occurring in distributed tree-based model learning.

(2 min) We study fast algorithms for computing fundamental properties of a positive semidefinite kernel matrix $K \in \mathbb{R}^{n \times n}$ corresponding to $n$ points $x_1,\ldots,x_n \in \mathbb{R}^d$. In particular, we consider estimating the sum of kernel matrix entries, along with its top eigenvalue and eigenvector. We show that the sum of matrix entries can be estimated to $1+\epsilon$ relative error in time $sublinear$ in $n$ and linear in $d$ for many popular kernels, including the Gaussian, exponential, and rational quadratic kernels. For these kernels, we also show that the top eigenvalue (and an approximate eigenvector) can be approximated to $1+\epsilon$ relative error in time $subquadratic$ in $n$ and linear in $d$. Our algorithms represent significant advances in the best known runtimes for these problems. They leverage the positive definiteness of the kernel matrix, along with a recent line of work on efficient kernel density estimation.

(2 min) Quantum dots (QDs) defined with electrostatic gates are a leading platform for a scalable quantum computing implementation. However, with increasing numbers of qubits, the complexity of the control parameter space also grows. Traditional measurement techniques, relying on complete or near-complete exploration via two-parameter scans (images) of the device response, quickly become impractical with increasing numbers of gates. Here we propose to circumvent this challenge by introducing a measurement technique relying on one-dimensional projections of the device response in the multidimensional parameter space. Dubbed the ``ray-based classification (RBC) framework,'' we use this machine learning approach to implement a classifier for QD states, enabling automated recognition of qubit-relevant parameter regimes. We show that RBC surpasses the 82 % accuracy benchmark from the experimental implementation of image-based classification techniques from prior work while reducing the number of measurement points needed by up to 70 %. The reduction in measurement cost is a significant gain for time-intensive QD measurements and is a step forward toward the scalability of these devices. We also discuss how the RBC-based optimizer, which tunes the device to a multiqubit regime, performs when tuning in the two-dimensional and three-dimensional parameter spaces defined by plunger and barrier gates that control the QDs.This work provides experimental validation of both efficient state identification and optimization with machine learning techniques for non-traditional measurements in quantum systems with high-dimensional parameter spaces and time-intensive measurements.

(2 min) Inspired by its success in natural language processing and computer vision, pre-training has attracted substantial attention in cheminformatics and bioinformatics, especially for molecule based tasks. A molecule can be represented by either a graph (where atoms are connected by bonds) or a SMILES sequence (where depth-first-search is applied to the molecular graph with specific rules). Existing works on molecule pre-training use either graph representations only or SMILES representations only. In this work, we propose to leverage both the representations and design a new pre-training algorithm, dual-view molecule pre-training (briefly, DMP), that can effectively combine the strengths of both types of molecule representations. The model of DMP consists of two branches: a Transformer branch that takes the SMILES sequence of a molecule as input, and a GNN branch that takes a molecular graph as input. The training of DMP contains three tasks: (1) predicting masked tokens in a SMILES sequence by the Transformer branch, (2) predicting masked atoms in a molecular graph by the GNN branch, and (3) maximizing the consistency between the two high-level representations output by the Transformer and GNN branches separately. After pre-training, we can use either the Transformer branch (this one is recommended according to empirical results), the GNN branch, or both for downstream tasks. DMP is tested on nine molecular property prediction tasks and achieves state-of-the-art performances on seven of them. Furthermore, we test DMP on three retrosynthesis tasks and achieve state-of-the-result on the USPTO-full dataset. Our code will be released soon.

(2 min) To evaluate clustering results is a significant part of cluster analysis. Since there are no true class labels for clustering in typical unsupervised learning, many internal cluster validity indices (CVIs), which use predicted labels and data, have been created. Without true labels, to design an effective CVI is as difficult as to create a clustering method. And it is crucial to have more CVIs because there are no universal CVIs that can be used to measure all datasets and no specific methods of selecting a proper CVI for clusters without true labels. Therefore, to apply a variety of CVIs to evaluate clustering results is necessary. In this paper, we propose a novel internal CVI -- the Distance-based Separability Index (DSI), based on a data separability measure. We compared the DSI with eight internal CVIs including studies from early Dunn (1974) to most recent CVDD (2019) and an external CVI as ground truth, by using clustering results of five clustering algorithms on 12 real and 97 synthetic datasets. Results show DSI is an effective, unique, and competitive CVI to other compared CVIs. We also summarized the general process to evaluate CVIs and created the rank-difference metric for comparison of CVIs' results.